diff --git a/metadata/trajectory_hera_001462.yaml b/metadata/trajectory_hera_001462.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f6b2ef2581399e933f2c37aa5c941e73ce2e1b65 --- /dev/null +++ b/metadata/trajectory_hera_001462.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_gateway_earth_orbit_00489_20261018T195529_022_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 30000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: gateway + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_gateway_earth_orbit_00489 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=26.0 + - env.viewpoint_radius_range=[36.88,40.57] + - env.orbit_speed_range=[0.336,0.433] + - env.camera_clipping_range=[0.05,120.0] + - env.lidar_max_distance=120.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=gateway + - env.spawn_offset_range=[0.2,1.5] + - env.spawn_attitude_noise_deg=15 + seed: 1167102211 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 7819.06698884815 + goal_position_error_m: + mean: 0.17008121998708361 + median_second_half: 0.004006450064480305 + final: 0.004113307222723961 + goal_orientation_error_deg: + mean: 2.147627858894756 + median_second_half: 0.4713082390534442 + final: 0.44835272030497186 +recorded_at: '2026-10-18T20:05:09' +trajectory_id: trajectory_hera_001462 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001463.yaml b/metadata/trajectory_hera_001463.yaml new file mode 100644 index 0000000000000000000000000000000000000000..5ca4ea8592a938fe14e1ce861156662adc9b6d09 --- /dev/null +++ b/metadata/trajectory_hera_001463.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_gateway_earth_orbit_00489_20261018T195529_022_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 30008333333 +end_ns: 60000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: gateway + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_gateway_earth_orbit_00489 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=26.0 + - env.viewpoint_radius_range=[36.88,40.57] + - env.orbit_speed_range=[0.336,0.433] + - env.camera_clipping_range=[0.05,120.0] + - env.lidar_max_distance=120.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=gateway + - env.spawn_offset_range=[0.2,1.5] + - env.spawn_attitude_noise_deg=15 + seed: 1167102211 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8189.133151769638 + goal_position_error_m: + mean: 0.11158753522328314 + median_second_half: 0.008683420717716217 + final: 0.008662878535687923 + goal_orientation_error_deg: + mean: 1.7547668518815425 + median_second_half: 0.31130548153676846 + final: 0.3129880292696923 +recorded_at: '2026-10-18T20:15:04' +trajectory_id: trajectory_hera_001463 +t0_sim_ns: 30000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001464.yaml b/metadata/trajectory_hera_001464.yaml new file mode 100644 index 0000000000000000000000000000000000000000..32c2cb3e0a94e614eed6035e7c7010c6a5a08839 --- /dev/null +++ b/metadata/trajectory_hera_001464.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_gateway_earth_orbit_00489_20261018T195529_022_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 60008333333 +end_ns: 90000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: gateway + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_gateway_earth_orbit_00489 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=26.0 + - env.viewpoint_radius_range=[36.88,40.57] + - env.orbit_speed_range=[0.336,0.433] + - env.camera_clipping_range=[0.05,120.0] + - env.lidar_max_distance=120.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=gateway + - env.spawn_offset_range=[0.2,1.5] + - env.spawn_attitude_noise_deg=15 + seed: 1167102211 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8159.366051435471 + goal_position_error_m: + mean: 0.11490564839140917 + median_second_half: 0.009710212238132954 + final: 0.009688882157206535 + goal_orientation_error_deg: + mean: 2.0787526833647294 + median_second_half: 0.5877386852140963 + final: 0.5953103767958786 +recorded_at: '2026-10-18T20:22:49' +trajectory_id: trajectory_hera_001464 +t0_sim_ns: 60000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001465.yaml b/metadata/trajectory_hera_001465.yaml new file mode 100644 index 0000000000000000000000000000000000000000..2563004d2ca2a3a731f7b8f79c9179b96622719f --- /dev/null +++ b/metadata/trajectory_hera_001465.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_iss_earth_approach_00488_20261018T195312_021_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 30000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: iss + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_iss_earth_approach_00488 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=70.0 + - env.viewpoint_radius_range=[85.82,94.41] + - env.orbit_speed_range=[0.64,0.7] + - env.camera_clipping_range=[0.05,250.0] + - env.lidar_max_distance=250.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=iss + - env.spawn_offset_range=[2.0,4.0] + - env.spawn_attitude_noise_deg=45 + seed: 274461408 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 5064.70565585047 + goal_position_error_m: + mean: 0.5018451526680608 + median_second_half: 0.01840542070567608 + final: 0.018153881654143333 + goal_orientation_error_deg: + mean: 18.932740792621505 + median_second_half: 0.6271655786720247 + final: 0.6031511671587888 +recorded_at: '2026-10-18T20:03:48' +trajectory_id: trajectory_hera_001465 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001466.yaml b/metadata/trajectory_hera_001466.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a9c922cbf929b26ebc35671e6f09abbd9f36ca48 --- /dev/null +++ b/metadata/trajectory_hera_001466.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_iss_earth_approach_00488_20261018T195312_021_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 30008333333 +end_ns: 60000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: iss + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_iss_earth_approach_00488 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=70.0 + - env.viewpoint_radius_range=[85.82,94.41] + - env.orbit_speed_range=[0.64,0.7] + - env.camera_clipping_range=[0.05,250.0] + - env.lidar_max_distance=250.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=iss + - env.spawn_offset_range=[2.0,4.0] + - env.spawn_attitude_noise_deg=45 + seed: 274461408 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 5006.826941238716 + goal_position_error_m: + mean: 0.5458328793122375 + median_second_half: 0.013915963470935822 + final: 0.014186822809278965 + goal_orientation_error_deg: + mean: 19.696649936373028 + median_second_half: 1.5286617566544978 + final: 1.5254383342530748 +recorded_at: '2026-10-18T20:15:21' +trajectory_id: trajectory_hera_001466 +t0_sim_ns: 30000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001467.yaml b/metadata/trajectory_hera_001467.yaml new file mode 100644 index 0000000000000000000000000000000000000000..0aa67f28185a13a08cd111244b3142ffa1e927b9 --- /dev/null +++ b/metadata/trajectory_hera_001467.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_iss_earth_approach_00488_20261018T195312_021_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 60008333333 +end_ns: 90000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: iss + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_iss_earth_approach_00488 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=70.0 + - env.viewpoint_radius_range=[85.82,94.41] + - env.orbit_speed_range=[0.64,0.7] + - env.camera_clipping_range=[0.05,250.0] + - env.lidar_max_distance=250.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=iss + - env.spawn_offset_range=[2.0,4.0] + - env.spawn_attitude_noise_deg=45 + seed: 274461408 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 3759.1622763574123 + goal_position_error_m: + mean: 0.8206574845984825 + median_second_half: 0.016030577942728996 + final: 0.003841773374006152 + goal_orientation_error_deg: + mean: 20.179301901556734 + median_second_half: 1.7620352333286702 + final: 1.051705911652805 +recorded_at: '2026-10-18T20:23:57' +trajectory_id: trajectory_hera_001467 +t0_sim_ns: 60000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001468.yaml b/metadata/trajectory_hera_001468.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b7c1921a39ee7a12b0fd9851e673b81b5f2b411e --- /dev/null +++ b/metadata/trajectory_hera_001468.yaml @@ -0,0 +1,147 @@ +trajectory_dir: zerog_satellite_mockup_floating_docking_00486_20261018T193101_019_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 40000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 19200 + imu: 4800 + joint_states: 4800 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: docking + target_kind: satellite_mockup + target_mount: floating + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_satellite_mockup_floating_docking_00486 + overrides: + - env.mode=docking + - env.goal_reached_pos_m=0.25 + - env.target_mount=floating + - env.target_kind=satellite_mockup + - env.target_scale=0.95 + - env.fill_light_intensity=54175 + - env.event_camera_preset=imx636_inspection + - env.docking_speed_range=[0.034,0.071] + seed: 64924280 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 10698.611259356141 + goal_position_error_m: + mean: 0.1310949411017918 + median_second_half: 0.030848165974020958 + final: 0.031031042337417603 + goal_orientation_error_deg: + mean: 5.030462693191474 + median_second_half: 0.590859174530293 + final: 0.590859174530293 +recorded_at: '2026-10-18T19:42:39' +trajectory_id: trajectory_hera_001468 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001469.yaml b/metadata/trajectory_hera_001469.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8b75efb9283b53173f3cb65821560b4d22da09a4 --- /dev/null +++ b/metadata/trajectory_hera_001469.yaml @@ -0,0 +1,147 @@ +trajectory_dir: zerog_satellite_mockup_floating_docking_00486_20261018T193101_019_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 40008333333 +end_ns: 80000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 19200 + imu: 4800 + joint_states: 4800 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: docking + target_kind: satellite_mockup + target_mount: floating + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_satellite_mockup_floating_docking_00486 + overrides: + - env.mode=docking + - env.goal_reached_pos_m=0.25 + - env.target_mount=floating + - env.target_kind=satellite_mockup + - env.target_scale=0.95 + - env.fill_light_intensity=54175 + - env.event_camera_preset=imx636_inspection + - env.docking_speed_range=[0.034,0.071] + seed: 64924280 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 10140.342026770115 + goal_position_error_m: + mean: 0.20316917768826456 + median_second_half: 0.019775988534092903 + final: 0.01969677023589611 + goal_orientation_error_deg: + mean: 13.69511926225287 + median_second_half: 0.39152013028389737 + final: 0.39147914914887727 +recorded_at: '2026-10-18T19:51:50' +trajectory_id: trajectory_hera_001469 +t0_sim_ns: 40000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001470.yaml b/metadata/trajectory_hera_001470.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f83746e4368867536699c4e1e7a55fe320666a7f --- /dev/null +++ b/metadata/trajectory_hera_001470.yaml @@ -0,0 +1,147 @@ +trajectory_dir: zerog_satellite_mockup_floating_docking_00486_20261018T193101_019_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 80008333333 +end_ns: 120000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 19200 + imu: 4800 + joint_states: 4800 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: docking + target_kind: satellite_mockup + target_mount: floating + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_satellite_mockup_floating_docking_00486 + overrides: + - env.mode=docking + - env.goal_reached_pos_m=0.25 + - env.target_mount=floating + - env.target_kind=satellite_mockup + - env.target_scale=0.95 + - env.fill_light_intensity=54175 + - env.event_camera_preset=imx636_inspection + - env.docking_speed_range=[0.034,0.071] + seed: 64924280 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 10799.015066493303 + goal_position_error_m: + mean: 0.12022731832560718 + median_second_half: 0.020403875038027763 + final: 0.0205445047467947 + goal_orientation_error_deg: + mean: 5.062139797835942 + median_second_half: 0.3240378346037449 + final: 0.3569183652681863 +recorded_at: '2026-10-18T20:02:10' +trajectory_id: trajectory_hera_001470 +t0_sim_ns: 80000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001471.yaml b/metadata/trajectory_hera_001471.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ed292f3d3c8f552b84e86bf8e03986fb69239a70 --- /dev/null +++ b/metadata/trajectory_hera_001471.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_cubesat_earth_orbit_00492_20261018T202406_025_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 30000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: cubesat + structure_scale: 1.7 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_cubesat_earth_orbit_00492 + overrides: + - env.structure_scale=1.7 + - env.structure_radius=0.765 + - env.viewpoint_radius_range=[1.59,1.75] + - env.orbit_speed_range=[0.082,0.108] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=cubesat + - env.spawn_offset_range=[0.2,0.46] + - env.spawn_attitude_noise_deg=15 + seed: 568935367 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8827.928630828857 + goal_position_error_m: + mean: 0.02603294664956753 + median_second_half: 0.013574011623859406 + final: 0.01061701774597168 + goal_orientation_error_deg: + mean: 1.7551108565803908 + median_second_half: 0.8050061489385557 + final: 0.6257212071059786 +recorded_at: '2026-10-18T20:30:47' +trajectory_id: trajectory_hera_001471 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001472.yaml b/metadata/trajectory_hera_001472.yaml new file mode 100644 index 0000000000000000000000000000000000000000..85d9306e2995faefaac37a6a4ba4a634b4514721 --- /dev/null +++ b/metadata/trajectory_hera_001472.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_cubesat_earth_orbit_00492_20261018T202406_025_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 30008333333 +end_ns: 60000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: cubesat + structure_scale: 1.7 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_cubesat_earth_orbit_00492 + overrides: + - env.structure_scale=1.7 + - env.structure_radius=0.765 + - env.viewpoint_radius_range=[1.59,1.75] + - env.orbit_speed_range=[0.082,0.108] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=cubesat + - env.spawn_offset_range=[0.2,0.46] + - env.spawn_attitude_noise_deg=15 + seed: 568935367 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8930.543828964233 + goal_position_error_m: + mean: 0.016598650529683626 + median_second_half: 0.006966960150748491 + final: 0.0067390440963208675 + goal_orientation_error_deg: + mean: 0.9570975238499935 + median_second_half: 0.773728418212393 + final: 0.5728092790706236 +recorded_at: '2026-10-18T20:36:59' +trajectory_id: trajectory_hera_001472 +t0_sim_ns: 30000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001473.yaml b/metadata/trajectory_hera_001473.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8860ce423e2b4995c3f134de587c38e169d00f4e --- /dev/null +++ b/metadata/trajectory_hera_001473.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_cubesat_earth_orbit_00492_20261018T202406_025_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 60008333333 +end_ns: 90000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: cubesat + structure_scale: 1.7 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_cubesat_earth_orbit_00492 + overrides: + - env.structure_scale=1.7 + - env.structure_radius=0.765 + - env.viewpoint_radius_range=[1.59,1.75] + - env.orbit_speed_range=[0.082,0.108] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=cubesat + - env.spawn_offset_range=[0.2,0.46] + - env.spawn_attitude_noise_deg=15 + seed: 568935367 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8770.782347679138 + goal_position_error_m: + mean: 0.024270487794547383 + median_second_half: 0.012207144871354103 + final: 0.014357347041368484 + goal_orientation_error_deg: + mean: 2.0193440144062254 + median_second_half: 0.6277906477027867 + final: 0.4353954245626345 +recorded_at: '2026-10-18T20:43:09' +trajectory_id: trajectory_hera_001473 +t0_sim_ns: 60000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001474.yaml b/metadata/trajectory_hera_001474.yaml new file mode 100644 index 0000000000000000000000000000000000000000..57f92f354ff4eb58b03ed6e85e50dd95fbfb510e --- /dev/null +++ b/metadata/trajectory_hera_001474.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_gateway_earth_approach_00491_20261018T202254_024_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 30000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: gateway + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_gateway_earth_approach_00491 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=26.0 + - env.viewpoint_radius_range=[38.88,42.76] + - env.orbit_speed_range=[0.214,0.418] + - env.camera_clipping_range=[0.05,120.0] + - env.lidar_max_distance=120.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=gateway + - env.spawn_offset_range=[2.0,4.0] + - env.spawn_attitude_noise_deg=45 + seed: 1983642780 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 5506.7667594999075 + goal_position_error_m: + mean: 0.5508446057057905 + median_second_half: 0.012561962939798832 + final: 0.011621439829468727 + goal_orientation_error_deg: + mean: 8.934044610984849 + median_second_half: 1.431727071623442 + final: 1.3859838006479857 +recorded_at: '2026-10-18T20:31:25' +trajectory_id: trajectory_hera_001474 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001475.yaml b/metadata/trajectory_hera_001475.yaml new file mode 100644 index 0000000000000000000000000000000000000000..72e0526c70773c066674c15a96ae9ad2a9bb2df6 --- /dev/null +++ b/metadata/trajectory_hera_001475.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_gateway_earth_approach_00491_20261018T202254_024_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 30008333333 +end_ns: 60000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: gateway + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_gateway_earth_approach_00491 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=26.0 + - env.viewpoint_radius_range=[38.88,42.76] + - env.orbit_speed_range=[0.214,0.418] + - env.camera_clipping_range=[0.05,120.0] + - env.lidar_max_distance=120.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=gateway + - env.spawn_offset_range=[2.0,4.0] + - env.spawn_attitude_noise_deg=45 + seed: 1983642780 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 5215.370587661862 + goal_position_error_m: + mean: 0.5232251965119024 + median_second_half: 0.008773665875196457 + final: 0.008829819038510323 + goal_orientation_error_deg: + mean: 12.516602755209906 + median_second_half: 0.513873817878405 + final: 0.46808022761867274 +recorded_at: '2026-10-18T20:40:17' +trajectory_id: trajectory_hera_001475 +t0_sim_ns: 30000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001476.yaml b/metadata/trajectory_hera_001476.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b9a5512cf96ecae0f19223a8feb49058487b4bef --- /dev/null +++ b/metadata/trajectory_hera_001476.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_gateway_earth_approach_00491_20261018T202254_024_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 60008333333 +end_ns: 90000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: gateway + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_gateway_earth_approach_00491 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=26.0 + - env.viewpoint_radius_range=[38.88,42.76] + - env.orbit_speed_range=[0.214,0.418] + - env.camera_clipping_range=[0.05,120.0] + - env.lidar_max_distance=120.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=gateway + - env.spawn_offset_range=[2.0,4.0] + - env.spawn_attitude_noise_deg=45 + seed: 1983642780 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 5532.177537634969 + goal_position_error_m: + mean: 0.45969617223534176 + median_second_half: 0.011056840419769287 + final: 0.011045500636100769 + goal_orientation_error_deg: + mean: 15.236248587359912 + median_second_half: 0.8589967666150967 + final: 0.8649830538966427 +recorded_at: '2026-10-18T20:48:17' +trajectory_id: trajectory_hera_001476 +t0_sim_ns: 60000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001477.yaml b/metadata/trajectory_hera_001477.yaml new file mode 100644 index 0000000000000000000000000000000000000000..c5e724166c572f6d62f8a2edb1ae8cfe061ba84b --- /dev/null +++ b/metadata/trajectory_hera_001477.yaml @@ -0,0 +1,137 @@ +trajectory_dir: orbital_satellite_mockup_black_orbit_00494_20261018T204314_027_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 30000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: satellite_mockup + structure_scale: 1.59 + viewpoint_mode: continuous + skydome: false + ambient_light_intensity: 1142.0 +scenario: + name: orbital_satellite_mockup_black_orbit_00494 + overrides: + - env.structure_scale=1.59 + - env.structure_radius=1.065 + - env.viewpoint_radius_range=[4.0,4.4] + - env.orbit_speed_range=[0.128,0.233] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=satellite_mockup + - env.spawn_offset_range=[0.2,0.7] + - env.spawn_attitude_noise_deg=15 + - env.skydome=false + - ++env.ambient_light_intensity=1142 + seed: 1286827665 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8826.622758865356 + goal_position_error_m: + mean: 0.02795534906290051 + median_second_half: 0.004884666297584772 + final: 0.004828366916626692 + goal_orientation_error_deg: + mean: 1.0979688513877848 + median_second_half: 0.7364383067476208 + final: 0.70195828451778 +recorded_at: '2026-10-18T20:50:38' +trajectory_id: trajectory_hera_001477 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001478.yaml b/metadata/trajectory_hera_001478.yaml new file mode 100644 index 0000000000000000000000000000000000000000..98428ad5f9c6bbc40c57904fc2f4eb97017dc620 --- /dev/null +++ b/metadata/trajectory_hera_001478.yaml @@ -0,0 +1,137 @@ +trajectory_dir: orbital_satellite_mockup_black_orbit_00494_20261018T204314_027_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 30008333333 +end_ns: 60000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: satellite_mockup + structure_scale: 1.59 + viewpoint_mode: continuous + skydome: false + ambient_light_intensity: 1142.0 +scenario: + name: orbital_satellite_mockup_black_orbit_00494 + overrides: + - env.structure_scale=1.59 + - env.structure_radius=1.065 + - env.viewpoint_radius_range=[4.0,4.4] + - env.orbit_speed_range=[0.128,0.233] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=satellite_mockup + - env.spawn_offset_range=[0.2,0.7] + - env.spawn_attitude_noise_deg=15 + - env.skydome=false + - ++env.ambient_light_intensity=1142 + seed: 1286827665 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8930.355924606323 + goal_position_error_m: + mean: 0.019237329468650873 + median_second_half: 0.008896030485630035 + final: 0.00882477592676878 + goal_orientation_error_deg: + mean: 0.5508998435128047 + median_second_half: 0.18273193286739892 + final: 0.20036061109456216 +recorded_at: '2026-10-18T20:55:35' +trajectory_id: trajectory_hera_001478 +t0_sim_ns: 30000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001479.yaml b/metadata/trajectory_hera_001479.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7a6254e7d6d69cf70a32d3fd0c9181730c2bd144 --- /dev/null +++ b/metadata/trajectory_hera_001479.yaml @@ -0,0 +1,137 @@ +trajectory_dir: orbital_satellite_mockup_black_orbit_00494_20261018T204314_027_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 60008333333 +end_ns: 90000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: satellite_mockup + structure_scale: 1.59 + viewpoint_mode: continuous + skydome: false + ambient_light_intensity: 1142.0 +scenario: + name: orbital_satellite_mockup_black_orbit_00494 + overrides: + - env.structure_scale=1.59 + - env.structure_radius=1.065 + - env.viewpoint_radius_range=[4.0,4.4] + - env.orbit_speed_range=[0.128,0.233] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=satellite_mockup + - env.spawn_offset_range=[0.2,0.7] + - env.spawn_attitude_noise_deg=15 + - env.skydome=false + - ++env.ambient_light_intensity=1142 + seed: 1286827665 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8939.206563949585 + goal_position_error_m: + mean: 0.011283152264487183 + median_second_half: 0.0017165779136121273 + final: 0.003141281893476844 + goal_orientation_error_deg: + mean: 0.7758068815798613 + median_second_half: 0.6156963570342511 + final: 0.6595904076527793 +recorded_at: '2026-10-18T21:00:23' +trajectory_id: trajectory_hera_001479 +t0_sim_ns: 60000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001480.yaml b/metadata/trajectory_hera_001480.yaml new file mode 100644 index 0000000000000000000000000000000000000000..603ece6230f5b7859e38f7c95b4c7d90b347d96d --- /dev/null +++ b/metadata/trajectory_hera_001480.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_venus_express_earth_approach_00493_20261018T203917_026_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 30000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: venus_express + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_venus_express_earth_approach_00493 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=4.4 + - env.viewpoint_radius_range=[10.16,11.17] + - env.orbit_speed_range=[0.184,0.223] + - env.camera_clipping_range=[0.05,80.0] + - env.lidar_max_distance=80.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=venus_express + - env.spawn_offset_range=[1.52,3.05] + - env.spawn_attitude_noise_deg=60 + seed: 1447722891 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 6222.371378138661 + goal_position_error_m: + mean: 0.4138856692737791 + median_second_half: 0.0047029112465679646 + final: 0.004707425832748413 + goal_orientation_error_deg: + mean: 5.186233814229605 + median_second_half: 0.5967832964182213 + final: 0.5940680827797297 +recorded_at: '2026-10-18T20:47:21' +trajectory_id: trajectory_hera_001480 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001481.yaml b/metadata/trajectory_hera_001481.yaml new file mode 100644 index 0000000000000000000000000000000000000000..26cdae0570fb4ffa63e2228219d6433d33740791 --- /dev/null +++ b/metadata/trajectory_hera_001481.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_venus_express_earth_approach_00493_20261018T203917_026_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 30008333333 +end_ns: 60000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: venus_express + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_venus_express_earth_approach_00493 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=4.4 + - env.viewpoint_radius_range=[10.16,11.17] + - env.orbit_speed_range=[0.184,0.223] + - env.camera_clipping_range=[0.05,80.0] + - env.lidar_max_distance=80.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=venus_express + - env.spawn_offset_range=[1.52,3.05] + - env.spawn_attitude_noise_deg=60 + seed: 1447722891 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 6698.496703535318 + goal_position_error_m: + mean: 0.27397885021484425 + median_second_half: 0.005683932453393936 + final: 0.005683922674506903 + goal_orientation_error_deg: + mean: 7.937383566404117 + median_second_half: 0.22989021359959827 + final: 0.18519985897338173 +recorded_at: '2026-10-18T20:52:31' +trajectory_id: trajectory_hera_001481 +t0_sim_ns: 30000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001482.yaml b/metadata/trajectory_hera_001482.yaml new file mode 100644 index 0000000000000000000000000000000000000000..11b12dfc35e1f6456c3420a39075bf9ca138b333 --- /dev/null +++ b/metadata/trajectory_hera_001482.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_venus_express_earth_approach_00493_20261018T203917_026_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 60008333333 +end_ns: 90000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: venus_express + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_venus_express_earth_approach_00493 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=4.4 + - env.viewpoint_radius_range=[10.16,11.17] + - env.orbit_speed_range=[0.184,0.223] + - env.camera_clipping_range=[0.05,80.0] + - env.lidar_max_distance=80.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=venus_express + - env.spawn_offset_range=[1.52,3.05] + - env.spawn_attitude_noise_deg=60 + seed: 1447722891 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 7463.378824263811 + goal_position_error_m: + mean: 0.2193847991405457 + median_second_half: 0.006326945964246988 + final: 0.0063200341537594795 + goal_orientation_error_deg: + mean: 2.524278653173249 + median_second_half: 0.2866771267305983 + final: 0.2666974360261359 +recorded_at: '2026-10-18T20:57:33' +trajectory_id: trajectory_hera_001482 +t0_sim_ns: 60000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001483.yaml b/metadata/trajectory_hera_001483.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1787cd73345c89de6d01a36815c062d6a991602c --- /dev/null +++ b/metadata/trajectory_hera_001483.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_venus_express_earth_orbit_00495_20261018T204824_028_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 30000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: venus_express + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_venus_express_earth_orbit_00495 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=4.4 + - env.viewpoint_radius_range=[9.51,10.46] + - env.orbit_speed_range=[0.125,0.247] + - env.camera_clipping_range=[0.05,80.0] + - env.lidar_max_distance=80.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=venus_express + - env.spawn_offset_range=[0.2,1.25] + - env.spawn_attitude_noise_deg=15 + seed: 1410607632 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8920.088437080383 + goal_position_error_m: + mean: 0.017187790406799793 + median_second_half: 0.009115812368690968 + final: 0.008780443109571934 + goal_orientation_error_deg: + mean: 1.0801701759870677 + median_second_half: 0.6680742496538756 + final: 0.6400463549513168 +recorded_at: '2026-10-18T20:54:03' +trajectory_id: trajectory_hera_001483 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001484.yaml b/metadata/trajectory_hera_001484.yaml new file mode 100644 index 0000000000000000000000000000000000000000..217986c438c010a7e604560fa96f43b9fbcb0504 --- /dev/null +++ b/metadata/trajectory_hera_001484.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_venus_express_earth_orbit_00495_20261018T204824_028_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 30008333333 +end_ns: 60000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: venus_express + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_venus_express_earth_orbit_00495 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=4.4 + - env.viewpoint_radius_range=[9.51,10.46] + - env.orbit_speed_range=[0.125,0.247] + - env.camera_clipping_range=[0.05,80.0] + - env.lidar_max_distance=80.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=venus_express + - env.spawn_offset_range=[0.2,1.25] + - env.spawn_attitude_noise_deg=15 + seed: 1410607632 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8613.186992645264 + goal_position_error_m: + mean: 0.05557677434738787 + median_second_half: 0.0030600123573094606 + final: 0.0027284868992865086 + goal_orientation_error_deg: + mean: 0.7762622419136403 + median_second_half: 0.45199561237096875 + final: 0.42387759118787743 +recorded_at: '2026-10-18T20:58:24' +trajectory_id: trajectory_hera_001484 +t0_sim_ns: 30000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001485.yaml b/metadata/trajectory_hera_001485.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ed872fe91d6f3708c2815209fbe479239bda2671 --- /dev/null +++ b/metadata/trajectory_hera_001485.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_venus_express_earth_orbit_00495_20261018T204824_028_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 60008333333 +end_ns: 90000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: venus_express + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_venus_express_earth_orbit_00495 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=4.4 + - env.viewpoint_radius_range=[9.51,10.46] + - env.orbit_speed_range=[0.125,0.247] + - env.camera_clipping_range=[0.05,80.0] + - env.lidar_max_distance=80.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=venus_express + - env.spawn_offset_range=[0.2,1.25] + - env.spawn_attitude_noise_deg=15 + seed: 1410607632 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8268.58251953125 + goal_position_error_m: + mean: 0.10132271915278165 + median_second_half: 0.005539378616958857 + final: 0.005527346394956112 + goal_orientation_error_deg: + mean: 1.3439558475067381 + median_second_half: 0.43632398344421247 + final: 0.4862363111755947 +recorded_at: '2026-10-18T21:03:49' +trajectory_id: trajectory_hera_001485 +t0_sim_ns: 60000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001486.yaml b/metadata/trajectory_hera_001486.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4937af7ec5e11f29556bdf3513f92a10f8a4dba9 --- /dev/null +++ b/metadata/trajectory_hera_001486.yaml @@ -0,0 +1,163 @@ +trajectory_dir: zerog_satellite_mockup_arm_ceiling_inspection_00490_20261018T200218_023_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 40000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 24000 + imu: 4800 + joint_states: 14400 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint + arm_ceiling: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint + arm_wall: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: inspection + target_kind: satellite_mockup + target_mount: arm_ceiling + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_satellite_mockup_arm_ceiling_inspection_00490 + overrides: + - env.mode=inspection + - env.goal_reached_pos_m=0.25 + - env.target_mount=arm_ceiling + - env.target_kind=satellite_mockup + - env.target_scale=0.88 + - env.fill_light_intensity=50183 + - env.event_camera_preset=imx636_inspection + - env.arc_speed_range=[0.035,0.04] + seed: 1583855416 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 9918.396093603224 + goal_position_error_m: + mean: 0.2501740885429326 + median_second_half: 0.01913287490606308 + final: 0.021048998460173607 + goal_orientation_error_deg: + mean: 5.121687103096327 + median_second_half: 0.7856493394698867 + final: 0.7471817140643748 +recorded_at: '2026-10-18T20:17:00' +trajectory_id: trajectory_hera_001486 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001487.yaml b/metadata/trajectory_hera_001487.yaml new file mode 100644 index 0000000000000000000000000000000000000000..68cf2c59e49d9e3cb04037a451d34772d15bd273 --- /dev/null +++ b/metadata/trajectory_hera_001487.yaml @@ -0,0 +1,163 @@ +trajectory_dir: zerog_satellite_mockup_arm_ceiling_inspection_00490_20261018T200218_023_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 40008333333 +end_ns: 80000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 24000 + imu: 4800 + joint_states: 14400 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint + arm_ceiling: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint + arm_wall: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: inspection + target_kind: satellite_mockup + target_mount: arm_ceiling + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_satellite_mockup_arm_ceiling_inspection_00490 + overrides: + - env.mode=inspection + - env.goal_reached_pos_m=0.25 + - env.target_mount=arm_ceiling + - env.target_kind=satellite_mockup + - env.target_scale=0.88 + - env.fill_light_intensity=50183 + - env.event_camera_preset=imx636_inspection + - env.arc_speed_range=[0.035,0.04] + seed: 1583855416 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 10923.497560679913 + goal_position_error_m: + mean: 0.1288331851706912 + median_second_half: 0.02554495260119438 + final: 0.02859165333211422 + goal_orientation_error_deg: + mean: 1.698681627692698 + median_second_half: 0.3625874222792969 + final: 1.4207887371110781 +recorded_at: '2026-10-18T20:27:30' +trajectory_id: trajectory_hera_001487 +t0_sim_ns: 40000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001488.yaml b/metadata/trajectory_hera_001488.yaml new file mode 100644 index 0000000000000000000000000000000000000000..c3988df38ca266318b1e700ef28327074b3cc806 --- /dev/null +++ b/metadata/trajectory_hera_001488.yaml @@ -0,0 +1,163 @@ +trajectory_dir: zerog_satellite_mockup_arm_ceiling_inspection_00490_20261018T200218_023_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 80008333333 +end_ns: 120000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 24000 + imu: 4800 + joint_states: 14400 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint + arm_ceiling: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint + arm_wall: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: inspection + target_kind: satellite_mockup + target_mount: arm_ceiling + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_satellite_mockup_arm_ceiling_inspection_00490 + overrides: + - env.mode=inspection + - env.goal_reached_pos_m=0.25 + - env.target_mount=arm_ceiling + - env.target_kind=satellite_mockup + - env.target_scale=0.88 + - env.fill_light_intensity=50183 + - env.event_camera_preset=imx636_inspection + - env.arc_speed_range=[0.035,0.04] + seed: 1583855416 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 10498.66739962995 + goal_position_error_m: + mean: 0.14295686667371643 + median_second_half: 0.032710444182157516 + final: 0.03161311894655228 + goal_orientation_error_deg: + mean: 10.743262129425016 + median_second_half: 0.5773558905411779 + final: 0.5521524925038308 +recorded_at: '2026-10-18T20:39:09' +trajectory_id: trajectory_hera_001488 +t0_sim_ns: 80000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001489.yaml b/metadata/trajectory_hera_001489.yaml new file mode 100644 index 0000000000000000000000000000000000000000..c471e32773d7c5c0126f6d45acda82cb60f2bc8f --- /dev/null +++ b/metadata/trajectory_hera_001489.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_cubesat_earth_orbit_00499_20261018T211113_032_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 30000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: cubesat + structure_scale: 2.05 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_cubesat_earth_orbit_00499 + overrides: + - env.structure_scale=2.05 + - env.structure_radius=0.922 + - env.viewpoint_radius_range=[2.47,2.72] + - env.orbit_speed_range=[0.156,0.166] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=cubesat + - env.spawn_offset_range=[0.2,0.55] + - env.spawn_attitude_noise_deg=15 + seed: 403503323 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8817.156466007233 + goal_position_error_m: + mean: 0.02961886729163263 + median_second_half: 0.007008788641542196 + final: 0.008125373162329197 + goal_orientation_error_deg: + mean: 1.0052889975813017 + median_second_half: 0.48052734024652183 + final: 0.34189397680656813 +recorded_at: '2026-10-18T21:20:03' +trajectory_id: trajectory_hera_001489 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001490.yaml b/metadata/trajectory_hera_001490.yaml new file mode 100644 index 0000000000000000000000000000000000000000..d121b523b10e456531948d24b36449a6195f4896 --- /dev/null +++ b/metadata/trajectory_hera_001490.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_cubesat_earth_orbit_00499_20261018T211113_032_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 30008333333 +end_ns: 60000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: cubesat + structure_scale: 2.05 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_cubesat_earth_orbit_00499 + overrides: + - env.structure_scale=2.05 + - env.structure_radius=0.922 + - env.viewpoint_radius_range=[2.47,2.72] + - env.orbit_speed_range=[0.156,0.166] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=cubesat + - env.spawn_offset_range=[0.2,0.55] + - env.spawn_attitude_noise_deg=15 + seed: 403503323 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8909.405097961426 + goal_position_error_m: + mean: 0.017620669055940504 + median_second_half: 0.008036873303353786 + final: 0.015491610392928123 + goal_orientation_error_deg: + mean: 1.2802713767522338 + median_second_half: 1.5428064380953594 + final: 1.5338632658725797 +recorded_at: '2026-10-18T21:27:17' +trajectory_id: trajectory_hera_001490 +t0_sim_ns: 30000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001491.yaml b/metadata/trajectory_hera_001491.yaml new file mode 100644 index 0000000000000000000000000000000000000000..bb2d9ccf1f1931b5bec4a81380c0ebd087882700 --- /dev/null +++ b/metadata/trajectory_hera_001491.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_cubesat_earth_orbit_00499_20261018T211113_032_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 60008333333 +end_ns: 90000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: cubesat + structure_scale: 2.05 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_cubesat_earth_orbit_00499 + overrides: + - env.structure_scale=2.05 + - env.structure_radius=0.922 + - env.viewpoint_radius_range=[2.47,2.72] + - env.orbit_speed_range=[0.156,0.166] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=cubesat + - env.spawn_offset_range=[0.2,0.55] + - env.spawn_attitude_noise_deg=15 + seed: 403503323 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8803.560083389282 + goal_position_error_m: + mean: 0.023446053947226153 + median_second_half: 0.014487236738204956 + final: 0.013730117119848728 + goal_orientation_error_deg: + mean: 2.7974672105244744 + median_second_half: 1.2086565084254106 + final: 0.9661135027904959 +recorded_at: '2026-10-18T21:33:22' +trajectory_id: trajectory_hera_001491 +t0_sim_ns: 60000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001492.yaml b/metadata/trajectory_hera_001492.yaml new file mode 100644 index 0000000000000000000000000000000000000000..e2e1979036b89aaaaaadc3c65a953d8cdb2470f7 --- /dev/null +++ b/metadata/trajectory_hera_001492.yaml @@ -0,0 +1,163 @@ +trajectory_dir: zerog_satellite_mockup_arm_wall_inspection_00496_20261018T205736_029_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 40000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 24000 + imu: 4800 + joint_states: 14400 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint + arm_ceiling: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint + arm_wall: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: inspection + target_kind: satellite_mockup + target_mount: arm_wall + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_satellite_mockup_arm_wall_inspection_00496 + overrides: + - env.mode=inspection + - env.goal_reached_pos_m=0.25 + - env.target_mount=arm_wall + - env.target_kind=satellite_mockup + - env.target_scale=0.94 + - env.fill_light_intensity=49662 + - env.event_camera_preset=imx636_inspection + - env.arc_speed_range=[0.06,0.061] + seed: 361715590 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 10806.193234734237 + goal_position_error_m: + mean: 0.13585866813749967 + median_second_half: 0.015712903812527657 + final: 0.017782682552933693 + goal_orientation_error_deg: + mean: 2.5404183082262337 + median_second_half: 1.1559821962679782 + final: 1.0766972536607642 +recorded_at: '2026-10-18T21:09:19' +trajectory_id: trajectory_hera_001492 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001493.yaml b/metadata/trajectory_hera_001493.yaml new file mode 100644 index 0000000000000000000000000000000000000000..5d30536f2af62c1eaa8f18299a828ccfb6d49c19 --- /dev/null +++ b/metadata/trajectory_hera_001493.yaml @@ -0,0 +1,163 @@ +trajectory_dir: zerog_satellite_mockup_arm_wall_inspection_00496_20261018T205736_029_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 40008333333 +end_ns: 80000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 24000 + imu: 4800 + joint_states: 14400 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint + arm_ceiling: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint + arm_wall: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: inspection + target_kind: satellite_mockup + target_mount: arm_wall + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_satellite_mockup_arm_wall_inspection_00496 + overrides: + - env.mode=inspection + - env.goal_reached_pos_m=0.25 + - env.target_mount=arm_wall + - env.target_kind=satellite_mockup + - env.target_scale=0.94 + - env.fill_light_intensity=49662 + - env.event_camera_preset=imx636_inspection + - env.arc_speed_range=[0.06,0.061] + seed: 361715590 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 10658.070123480633 + goal_position_error_m: + mean: 0.12718200014858852 + median_second_half: 0.025053491815924644 + final: 0.010228296741843224 + goal_orientation_error_deg: + mean: 7.8568714143415495 + median_second_half: 1.0538435474196073 + final: 1.228860314712014 +recorded_at: '2026-10-18T21:22:34' +trajectory_id: trajectory_hera_001493 +t0_sim_ns: 40000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001494.yaml b/metadata/trajectory_hera_001494.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ad4d0998fa476b0c4a23e6da1d56391f6f159545 --- /dev/null +++ b/metadata/trajectory_hera_001494.yaml @@ -0,0 +1,163 @@ +trajectory_dir: zerog_satellite_mockup_arm_wall_inspection_00496_20261018T205736_029_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 80008333333 +end_ns: 120000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 24000 + imu: 4800 + joint_states: 14400 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint + arm_ceiling: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint + arm_wall: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: inspection + target_kind: satellite_mockup + target_mount: arm_wall + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_satellite_mockup_arm_wall_inspection_00496 + overrides: + - env.mode=inspection + - env.goal_reached_pos_m=0.25 + - env.target_mount=arm_wall + - env.target_kind=satellite_mockup + - env.target_scale=0.94 + - env.fill_light_intensity=49662 + - env.event_camera_preset=imx636_inspection + - env.arc_speed_range=[0.06,0.061] + seed: 361715590 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 10425.347849942744 + goal_position_error_m: + mean: 0.17198844264235877 + median_second_half: 0.025690199807286263 + final: 0.007284505758434534 + goal_orientation_error_deg: + mean: 5.92072449311153 + median_second_half: 0.9962650527163999 + final: 1.3871704392945696 +recorded_at: '2026-10-18T21:34:20' +trajectory_id: trajectory_hera_001494 +t0_sim_ns: 80000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001495.yaml b/metadata/trajectory_hera_001495.yaml new file mode 100644 index 0000000000000000000000000000000000000000..60ba4057f5548b2b671ae5e4b0bf256145b24445 --- /dev/null +++ b/metadata/trajectory_hera_001495.yaml @@ -0,0 +1,147 @@ +trajectory_dir: zerog_satellite_mockup_floating_docking_00498_20261018T210352_031_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 40000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 19200 + imu: 4800 + joint_states: 4800 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: docking + target_kind: satellite_mockup + target_mount: floating + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_satellite_mockup_floating_docking_00498 + overrides: + - env.mode=docking + - env.goal_reached_pos_m=0.25 + - env.target_mount=floating + - env.target_kind=satellite_mockup + - env.target_scale=1.11 + - env.fill_light_intensity=52027 + - env.event_camera_preset=imx636_inspection + - env.docking_speed_range=[0.04,0.06] + seed: 790313805 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 9448.103204205632 + goal_position_error_m: + mean: 0.2134594158873315 + median_second_half: 0.07349090278148651 + final: 0.07334276288747787 + goal_orientation_error_deg: + mean: 5.43161400603391 + median_second_half: 3.787695064608911 + final: 3.7894576803015974 +recorded_at: '2026-10-18T21:17:06' +trajectory_id: trajectory_hera_001495 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001496.yaml b/metadata/trajectory_hera_001496.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1433422de26a14078c516f58abea547ea489f120 --- /dev/null +++ b/metadata/trajectory_hera_001496.yaml @@ -0,0 +1,147 @@ +trajectory_dir: zerog_satellite_mockup_floating_docking_00498_20261018T210352_031_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 40008333333 +end_ns: 80000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 19200 + imu: 4800 + joint_states: 4800 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: docking + target_kind: satellite_mockup + target_mount: floating + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_satellite_mockup_floating_docking_00498 + overrides: + - env.mode=docking + - env.goal_reached_pos_m=0.25 + - env.target_mount=floating + - env.target_kind=satellite_mockup + - env.target_scale=1.11 + - env.fill_light_intensity=52027 + - env.event_camera_preset=imx636_inspection + - env.docking_speed_range=[0.04,0.06] + seed: 790313805 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 8191.320025265217 + goal_position_error_m: + mean: 0.2742327091973879 + median_second_half: 0.0803074762225151 + final: 0.04367843270301819 + goal_orientation_error_deg: + mean: 21.80553474123912 + median_second_half: 2.0002618795732543 + final: 2.682000136105152 +recorded_at: '2026-10-18T21:30:17' +trajectory_id: trajectory_hera_001496 +t0_sim_ns: 40000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001497.yaml b/metadata/trajectory_hera_001497.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f2c9dbef09d5627b8d16c1e46c032914f1ce14dc --- /dev/null +++ b/metadata/trajectory_hera_001497.yaml @@ -0,0 +1,147 @@ +trajectory_dir: zerog_satellite_mockup_floating_docking_00498_20261018T210352_031_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 80008333333 +end_ns: 120000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 19200 + imu: 4800 + joint_states: 4800 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: docking + target_kind: satellite_mockup + target_mount: floating + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_satellite_mockup_floating_docking_00498 + overrides: + - env.mode=docking + - env.goal_reached_pos_m=0.25 + - env.target_mount=floating + - env.target_kind=satellite_mockup + - env.target_scale=1.11 + - env.fill_light_intensity=52027 + - env.event_camera_preset=imx636_inspection + - env.docking_speed_range=[0.04,0.06] + seed: 790313805 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 9876.379112022376 + goal_position_error_m: + mean: 0.12538974625441685 + median_second_half: 0.028704406693577766 + final: 0.02826080098748207 + goal_orientation_error_deg: + mean: 6.104999508032717 + median_second_half: 0.9969890527684212 + final: 1.1394394780982071 +recorded_at: '2026-10-18T21:40:00' +trajectory_id: trajectory_hera_001497 +t0_sim_ns: 80000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001498.yaml b/metadata/trajectory_hera_001498.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1abdd2f1df3ff2c1285512299eaeec95a8f4a298 --- /dev/null +++ b/metadata/trajectory_hera_001498.yaml @@ -0,0 +1,137 @@ +trajectory_dir: orbital_cubesat_black_orbit_00500_20261018T213325_033_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 30000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: cubesat + structure_scale: 2.81 + viewpoint_mode: continuous + skydome: false + ambient_light_intensity: 1472.0 +scenario: + name: orbital_cubesat_black_orbit_00500 + overrides: + - env.structure_scale=2.81 + - env.structure_radius=1.264 + - env.viewpoint_radius_range=[2.97,3.27] + - env.orbit_speed_range=[0.074,0.15] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=cubesat + - env.spawn_offset_range=[0.2,0.6] + - env.spawn_attitude_noise_deg=15 + - env.skydome=false + - ++env.ambient_light_intensity=1472 + seed: 213491634 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8811.086492538452 + goal_position_error_m: + mean: 0.0308909653067168 + median_second_half: 0.00992095097899437 + final: 0.011560529470443726 + goal_orientation_error_deg: + mean: 1.106324453874852 + median_second_half: 0.6635791847672139 + final: 0.936806761887558 +recorded_at: '2026-10-18T21:40:41' +trajectory_id: trajectory_hera_001498 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001499.yaml b/metadata/trajectory_hera_001499.yaml new file mode 100644 index 0000000000000000000000000000000000000000..10cf4b57ce1490a4eaa0e7efaed59169b6eeea41 --- /dev/null +++ b/metadata/trajectory_hera_001499.yaml @@ -0,0 +1,137 @@ +trajectory_dir: orbital_cubesat_black_orbit_00500_20261018T213325_033_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 30008333333 +end_ns: 60000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: cubesat + structure_scale: 2.81 + viewpoint_mode: continuous + skydome: false + ambient_light_intensity: 1472.0 +scenario: + name: orbital_cubesat_black_orbit_00500 + overrides: + - env.structure_scale=2.81 + - env.structure_radius=1.264 + - env.viewpoint_radius_range=[2.97,3.27] + - env.orbit_speed_range=[0.074,0.15] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=cubesat + - env.spawn_offset_range=[0.2,0.6] + - env.spawn_attitude_noise_deg=15 + - env.skydome=false + - ++env.ambient_light_intensity=1472 + seed: 213491634 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8877.332372665405 + goal_position_error_m: + mean: 0.020491455053520756 + median_second_half: 0.0024337461218237877 + final: 0.003067707410082221 + goal_orientation_error_deg: + mean: 0.5368571810644859 + median_second_half: 0.3174985154428395 + final: 0.3834974861343356 +recorded_at: '2026-10-18T21:47:41' +trajectory_id: trajectory_hera_001499 +t0_sim_ns: 30000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001500.yaml b/metadata/trajectory_hera_001500.yaml new file mode 100644 index 0000000000000000000000000000000000000000..20ace71c66b60a689afcfe24521052368ab01ea8 --- /dev/null +++ b/metadata/trajectory_hera_001500.yaml @@ -0,0 +1,137 @@ +trajectory_dir: orbital_cubesat_black_orbit_00500_20261018T213325_033_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 60008333333 +end_ns: 90000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: cubesat + structure_scale: 2.81 + viewpoint_mode: continuous + skydome: false + ambient_light_intensity: 1472.0 +scenario: + name: orbital_cubesat_black_orbit_00500 + overrides: + - env.structure_scale=2.81 + - env.structure_radius=1.264 + - env.viewpoint_radius_range=[2.97,3.27] + - env.orbit_speed_range=[0.074,0.15] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=cubesat + - env.spawn_offset_range=[0.2,0.6] + - env.spawn_attitude_noise_deg=15 + - env.skydome=false + - ++env.ambient_light_intensity=1472 + seed: 213491634 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8852.66969871521 + goal_position_error_m: + mean: 0.024697010430887935 + median_second_half: 0.00839821994304657 + final: 0.009439547546207905 + goal_orientation_error_deg: + mean: 1.0002597993768132 + median_second_half: 0.6762217581992017 + final: 0.8319801667965228 +recorded_at: '2026-10-18T21:56:01' +trajectory_id: trajectory_hera_001500 +t0_sim_ns: 60000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001501.yaml b/metadata/trajectory_hera_001501.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7585894753cad905a76288d4696a22fb828ebc84 --- /dev/null +++ b/metadata/trajectory_hera_001501.yaml @@ -0,0 +1,163 @@ +trajectory_dir: zerog_cubesat_arm_wall_inspection_00501_20261018T213425_034_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 40000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 24000 + imu: 4800 + joint_states: 14400 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint + arm_ceiling: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint + arm_wall: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: inspection + target_kind: cubesat + target_mount: arm_wall + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_cubesat_arm_wall_inspection_00501 + overrides: + - env.mode=inspection + - env.goal_reached_pos_m=0.25 + - env.target_mount=arm_wall + - env.target_kind=cubesat + - env.target_scale=2.43 + - env.fill_light_intensity=31673 + - env.event_camera_preset=imx636_inspection + - env.arc_speed_range=[0.071,0.099] + seed: 1246800582 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 10390.308321163058 + goal_position_error_m: + mean: 0.16508960772649456 + median_second_half: 0.005157266743481159 + final: 0.23942779004573822 + goal_orientation_error_deg: + mean: 5.947717094568498 + median_second_half: 1.5852322648465993 + final: 1.8796405253874175 +recorded_at: '2026-10-18T21:46:53' +trajectory_id: trajectory_hera_001501 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001502.yaml b/metadata/trajectory_hera_001502.yaml new file mode 100644 index 0000000000000000000000000000000000000000..86c63feac63aa722a09b9de191afcd85f3f65621 --- /dev/null +++ b/metadata/trajectory_hera_001502.yaml @@ -0,0 +1,163 @@ +trajectory_dir: zerog_cubesat_arm_wall_inspection_00501_20261018T213425_034_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 40008333333 +end_ns: 80000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 24000 + imu: 4800 + joint_states: 14400 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint + arm_ceiling: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint + arm_wall: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: inspection + target_kind: cubesat + target_mount: arm_wall + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_cubesat_arm_wall_inspection_00501 + overrides: + - env.mode=inspection + - env.goal_reached_pos_m=0.25 + - env.target_mount=arm_wall + - env.target_kind=cubesat + - env.target_scale=2.43 + - env.fill_light_intensity=31673 + - env.event_camera_preset=imx636_inspection + - env.arc_speed_range=[0.071,0.099] + seed: 1246800582 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 10154.345197990537 + goal_position_error_m: + mean: 0.18172635650146113 + median_second_half: 0.008512686938047409 + final: 0.024727605283260345 + goal_orientation_error_deg: + mean: 9.850059621979986 + median_second_half: 1.3333211211564833 + final: 1.2907281682139895 +recorded_at: '2026-10-18T22:01:11' +trajectory_id: trajectory_hera_001502 +t0_sim_ns: 40000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001503.yaml b/metadata/trajectory_hera_001503.yaml new file mode 100644 index 0000000000000000000000000000000000000000..110151547f7fee2bc428f3904d819a9ecab02c5c --- /dev/null +++ b/metadata/trajectory_hera_001503.yaml @@ -0,0 +1,163 @@ +trajectory_dir: zerog_cubesat_arm_wall_inspection_00501_20261018T213425_034_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 80008333333 +end_ns: 120000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 24000 + imu: 4800 + joint_states: 14400 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint + arm_ceiling: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint + arm_wall: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: inspection + target_kind: cubesat + target_mount: arm_wall + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_cubesat_arm_wall_inspection_00501 + overrides: + - env.mode=inspection + - env.goal_reached_pos_m=0.25 + - env.target_mount=arm_wall + - env.target_kind=cubesat + - env.target_scale=2.43 + - env.fill_light_intensity=31673 + - env.event_camera_preset=imx636_inspection + - env.arc_speed_range=[0.071,0.099] + seed: 1246800582 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 10867.842076539993 + goal_position_error_m: + mean: 0.11219339839942252 + median_second_half: 0.011278321966528893 + final: 0.001347105368040502 + goal_orientation_error_deg: + mean: 2.820835305300634 + median_second_half: 1.7196777084691304 + final: 1.6775491016684918 +recorded_at: '2026-10-18T22:14:44' +trajectory_id: trajectory_hera_001503 +t0_sim_ns: 80000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001504.yaml b/metadata/trajectory_hera_001504.yaml new file mode 100644 index 0000000000000000000000000000000000000000..870c8755a1c7f7c790e4845df1699a682dfdc7f1 --- /dev/null +++ b/metadata/trajectory_hera_001504.yaml @@ -0,0 +1,163 @@ +trajectory_dir: zerog_satellite_mockup_arm_wall_docking_00502_20261018T214006_035_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 40000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 24000 + imu: 4800 + joint_states: 14400 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint + arm_ceiling: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint + arm_wall: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: docking + target_kind: satellite_mockup + target_mount: arm_wall + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_satellite_mockup_arm_wall_docking_00502 + overrides: + - env.mode=docking + - env.goal_reached_pos_m=0.25 + - env.target_mount=arm_wall + - env.target_kind=satellite_mockup + - env.target_scale=1.15 + - env.fill_light_intensity=39442 + - env.event_camera_preset=imx636_inspection + - env.docking_speed_range=[0.048,0.071] + seed: 1874353690 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 9957.152748793364 + goal_position_error_m: + mean: 0.2824503868334042 + median_second_half: 0.020984644070267677 + final: 0.020482761785387993 + goal_orientation_error_deg: + mean: 1.3824441097772235 + median_second_half: 0.25482072423526214 + final: 0.2526760181554517 +recorded_at: '2026-10-18T21:53:34' +trajectory_id: trajectory_hera_001504 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001505.yaml b/metadata/trajectory_hera_001505.yaml new file mode 100644 index 0000000000000000000000000000000000000000..dd412e3dfbf7eadf0824f7428e2eb199d547bced --- /dev/null +++ b/metadata/trajectory_hera_001505.yaml @@ -0,0 +1,163 @@ +trajectory_dir: zerog_satellite_mockup_arm_wall_docking_00502_20261018T214006_035_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 40008333333 +end_ns: 80000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 24000 + imu: 4800 + joint_states: 14400 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint + arm_ceiling: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint + arm_wall: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: docking + target_kind: satellite_mockup + target_mount: arm_wall + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_satellite_mockup_arm_wall_docking_00502 + overrides: + - env.mode=docking + - env.goal_reached_pos_m=0.25 + - env.target_mount=arm_wall + - env.target_kind=satellite_mockup + - env.target_scale=1.15 + - env.fill_light_intensity=39442 + - env.event_camera_preset=imx636_inspection + - env.docking_speed_range=[0.048,0.071] + seed: 1874353690 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 11000.630892693996 + goal_position_error_m: + mean: 0.10939824645653683 + median_second_half: 0.020112887024879456 + final: 0.019980093464255333 + goal_orientation_error_deg: + mean: 2.1989874363810964 + median_second_half: 0.3680652339936471 + final: 0.28869840915767253 +recorded_at: '2026-10-18T22:06:25' +trajectory_id: trajectory_hera_001505 +t0_sim_ns: 40000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001506.yaml b/metadata/trajectory_hera_001506.yaml new file mode 100644 index 0000000000000000000000000000000000000000..706749575c761ba369939407a92c5bb3af27b273 --- /dev/null +++ b/metadata/trajectory_hera_001506.yaml @@ -0,0 +1,163 @@ +trajectory_dir: zerog_satellite_mockup_arm_wall_docking_00502_20261018T214006_035_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 80008333333 +end_ns: 120000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 24000 + imu: 4800 + joint_states: 14400 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint + arm_ceiling: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint + arm_wall: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: docking + target_kind: satellite_mockup + target_mount: arm_wall + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_satellite_mockup_arm_wall_docking_00502 + overrides: + - env.mode=docking + - env.goal_reached_pos_m=0.25 + - env.target_mount=arm_wall + - env.target_kind=satellite_mockup + - env.target_scale=1.15 + - env.fill_light_intensity=39442 + - env.event_camera_preset=imx636_inspection + - env.docking_speed_range=[0.048,0.071] + seed: 1874353690 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 10102.928583577275 + goal_position_error_m: + mean: 0.25252920267345763 + median_second_half: 0.021686704829335213 + final: 0.021686704829335213 + goal_orientation_error_deg: + mean: 2.8009874170131095 + median_second_half: 0.40695633112769974 + final: 0.40695633112769974 +recorded_at: '2026-10-18T22:18:01' +trajectory_id: trajectory_hera_001506 +t0_sim_ns: 80000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001507.yaml b/metadata/trajectory_hera_001507.yaml new file mode 100644 index 0000000000000000000000000000000000000000..476e525cff73320e5a3cc929e5ec9c3320d4cdf3 --- /dev/null +++ b/metadata/trajectory_hera_001507.yaml @@ -0,0 +1,137 @@ +trajectory_dir: orbital_iss_black_orbit_00503_20261018T215606_036_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 30000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: iss + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: false + ambient_light_intensity: 587.0 +scenario: + name: orbital_iss_black_orbit_00503 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=70.0 + - env.viewpoint_radius_range=[81.62,89.79] + - env.orbit_speed_range=[0.46,0.524] + - env.camera_clipping_range=[0.05,250.0] + - env.lidar_max_distance=250.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=iss + - env.spawn_offset_range=[0.2,1.5] + - env.spawn_attitude_noise_deg=15 + - env.skydome=false + - ++env.ambient_light_intensity=587 + seed: 1462860262 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8828.28744316101 + goal_position_error_m: + mean: 0.03319514757393553 + median_second_half: 0.015414301306009293 + final: 0.014998695813119411 + goal_orientation_error_deg: + mean: 1.1922625968328868 + median_second_half: 0.8558244639107431 + final: 0.8414105236061947 +recorded_at: '2026-10-18T22:07:23' +trajectory_id: trajectory_hera_001507 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001508.yaml b/metadata/trajectory_hera_001508.yaml new file mode 100644 index 0000000000000000000000000000000000000000..bc3cf2d79dff4ce7815fedcd247a2b7a715461da --- /dev/null +++ b/metadata/trajectory_hera_001508.yaml @@ -0,0 +1,137 @@ +trajectory_dir: orbital_iss_black_orbit_00503_20261018T215606_036_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 30008333333 +end_ns: 60000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: iss + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: false + ambient_light_intensity: 587.0 +scenario: + name: orbital_iss_black_orbit_00503 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=70.0 + - env.viewpoint_radius_range=[81.62,89.79] + - env.orbit_speed_range=[0.46,0.524] + - env.camera_clipping_range=[0.05,250.0] + - env.lidar_max_distance=250.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=iss + - env.spawn_offset_range=[0.2,1.5] + - env.spawn_attitude_noise_deg=15 + - env.skydome=false + - ++env.ambient_light_intensity=587 + seed: 1462860262 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8739.38855957985 + goal_position_error_m: + mean: 0.041902420918559315 + median_second_half: 0.011628391221165657 + final: 0.01131043117493391 + goal_orientation_error_deg: + mean: 0.9808403843761293 + median_second_half: 0.29075960550270896 + final: 0.263694473988431 +recorded_at: '2026-10-18T22:16:06' +trajectory_id: trajectory_hera_001508 +t0_sim_ns: 30000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001509.yaml b/metadata/trajectory_hera_001509.yaml new file mode 100644 index 0000000000000000000000000000000000000000..22c69cf4247c79fcf56ed237069f3e7f4e945165 --- /dev/null +++ b/metadata/trajectory_hera_001509.yaml @@ -0,0 +1,137 @@ +trajectory_dir: orbital_iss_black_orbit_00503_20261018T215606_036_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 60008333333 +end_ns: 90000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: iss + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: false + ambient_light_intensity: 587.0 +scenario: + name: orbital_iss_black_orbit_00503 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=70.0 + - env.viewpoint_radius_range=[81.62,89.79] + - env.orbit_speed_range=[0.46,0.524] + - env.camera_clipping_range=[0.05,250.0] + - env.lidar_max_distance=250.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=iss + - env.spawn_offset_range=[0.2,1.5] + - env.spawn_attitude_noise_deg=15 + - env.skydome=false + - ++env.ambient_light_intensity=587 + seed: 1462860262 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8107.280872583389 + goal_position_error_m: + mean: 0.12100858436866456 + median_second_half: 0.014271078631281853 + final: 0.014130537398159504 + goal_orientation_error_deg: + mean: 2.7855274052422914 + median_second_half: 1.4407881713311101 + final: 1.4322821312152114 +recorded_at: '2026-10-18T22:24:44' +trajectory_id: trajectory_hera_001509 +t0_sim_ns: 60000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001510.yaml b/metadata/trajectory_hera_001510.yaml new file mode 100644 index 0000000000000000000000000000000000000000..5ea10036dab98462f6cded8a0b445993354d2dec --- /dev/null +++ b/metadata/trajectory_hera_001510.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_iss_earth_approach_00505_20261018T221807_038_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 30000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: iss + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_iss_earth_approach_00505 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=70.0 + - env.viewpoint_radius_range=[75.46,83.01] + - env.orbit_speed_range=[0.444,0.661] + - env.camera_clipping_range=[0.05,250.0] + - env.lidar_max_distance=250.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=iss + - env.spawn_offset_range=[2.0,4.0] + - env.spawn_attitude_noise_deg=45 + seed: 1830920278 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 3245.361950159073 + goal_position_error_m: + mean: 0.9051236852483694 + median_second_half: 0.10091777890920639 + final: 0.01784353516995907 + goal_orientation_error_deg: + mean: 18.691456203071457 + median_second_half: 4.22216040950566 + final: 1.4445617441258365 +recorded_at: '2026-10-18T22:29:17' +trajectory_id: trajectory_hera_001510 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001511.yaml b/metadata/trajectory_hera_001511.yaml new file mode 100644 index 0000000000000000000000000000000000000000..59314597db42e5ab4a7f115057fcbfaed1dcbbf4 --- /dev/null +++ b/metadata/trajectory_hera_001511.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_iss_earth_approach_00505_20261018T221807_038_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 30008333333 +end_ns: 60000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: iss + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_iss_earth_approach_00505 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=70.0 + - env.viewpoint_radius_range=[75.46,83.01] + - env.orbit_speed_range=[0.444,0.661] + - env.camera_clipping_range=[0.05,250.0] + - env.lidar_max_distance=250.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=iss + - env.spawn_offset_range=[2.0,4.0] + - env.spawn_attitude_noise_deg=45 + seed: 1830920278 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 4166.102608174086 + goal_position_error_m: + mean: 0.7303353918877205 + median_second_half: 0.019690576940774918 + final: 0.017730016261339188 + goal_orientation_error_deg: + mean: 23.256668033261708 + median_second_half: 0.9172631360729185 + final: 0.9000744314180525 +recorded_at: '2026-10-18T22:39:26' +trajectory_id: trajectory_hera_001511 +t0_sim_ns: 30000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001512.yaml b/metadata/trajectory_hera_001512.yaml new file mode 100644 index 0000000000000000000000000000000000000000..bc92b909f48d41c955eeeedbc84891e2abcf499f --- /dev/null +++ b/metadata/trajectory_hera_001512.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_iss_earth_approach_00505_20261018T221807_038_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 60008333333 +end_ns: 90000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: iss + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_iss_earth_approach_00505 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=70.0 + - env.viewpoint_radius_range=[75.46,83.01] + - env.orbit_speed_range=[0.444,0.661] + - env.camera_clipping_range=[0.05,250.0] + - env.lidar_max_distance=250.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=iss + - env.spawn_offset_range=[2.0,4.0] + - env.spawn_attitude_noise_deg=45 + seed: 1830920278 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 4192.0761263147 + goal_position_error_m: + mean: 0.6513343575469333 + median_second_half: 0.020076939836144447 + final: 0.016294194385409355 + goal_orientation_error_deg: + mean: 22.405866134810303 + median_second_half: 1.6289178346128836 + final: 0.9652836348063394 +recorded_at: '2026-10-18T22:48:15' +trajectory_id: trajectory_hera_001512 +t0_sim_ns: 60000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001513.yaml b/metadata/trajectory_hera_001513.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a1a0d02fdc2f95c45faf1e73f7bcdb331a190a0b --- /dev/null +++ b/metadata/trajectory_hera_001513.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_satellite_mockup_earth_orbit_00506_20261018T222452_039_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 30000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: satellite_mockup + structure_scale: 1.49 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_satellite_mockup_earth_orbit_00506 + overrides: + - env.structure_scale=1.49 + - env.structure_radius=0.998 + - env.viewpoint_radius_range=[2.65,2.91] + - env.orbit_speed_range=[0.074,0.17] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=satellite_mockup + - env.spawn_offset_range=[0.2,0.56] + - env.spawn_attitude_noise_deg=15 + seed: 444853753 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8931.910059928894 + goal_position_error_m: + mean: 0.015695453626088228 + median_second_half: 0.005957204848527908 + final: 0.0039959014393389225 + goal_orientation_error_deg: + mean: 0.8034996721206505 + median_second_half: 0.822954552056027 + final: 1.0995413549484008 +recorded_at: '2026-10-18T22:33:06' +trajectory_id: trajectory_hera_001513 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001514.yaml b/metadata/trajectory_hera_001514.yaml new file mode 100644 index 0000000000000000000000000000000000000000..de21de2a9987b2fdd9a4b886ac424d79073e2669 --- /dev/null +++ b/metadata/trajectory_hera_001514.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_satellite_mockup_earth_orbit_00506_20261018T222452_039_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 30008333333 +end_ns: 60000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: satellite_mockup + structure_scale: 1.49 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_satellite_mockup_earth_orbit_00506 + overrides: + - env.structure_scale=1.49 + - env.structure_radius=0.998 + - env.viewpoint_radius_range=[2.65,2.91] + - env.orbit_speed_range=[0.074,0.17] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=satellite_mockup + - env.spawn_offset_range=[0.2,0.56] + - env.spawn_attitude_noise_deg=15 + seed: 444853753 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8784.093774080276 + goal_position_error_m: + mean: 0.030656525569711116 + median_second_half: 0.004990528337657452 + final: 0.00524544483050704 + goal_orientation_error_deg: + mean: 1.0398879926191842 + median_second_half: 0.5534387025019089 + final: 0.4681627234972418 +recorded_at: '2026-10-18T22:39:38' +trajectory_id: trajectory_hera_001514 +t0_sim_ns: 30000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001515.yaml b/metadata/trajectory_hera_001515.yaml new file mode 100644 index 0000000000000000000000000000000000000000..d8489a97429683cf0c5cec22147b964e8c78c4f2 --- /dev/null +++ b/metadata/trajectory_hera_001515.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_satellite_mockup_earth_orbit_00506_20261018T222452_039_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 60008333333 +end_ns: 90000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: satellite_mockup + structure_scale: 1.49 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_satellite_mockup_earth_orbit_00506 + overrides: + - env.structure_scale=1.49 + - env.structure_radius=0.998 + - env.viewpoint_radius_range=[2.65,2.91] + - env.orbit_speed_range=[0.074,0.17] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=satellite_mockup + - env.spawn_offset_range=[0.2,0.56] + - env.spawn_attitude_noise_deg=15 + seed: 444853753 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8891.604311943054 + goal_position_error_m: + mean: 0.018209033574525305 + median_second_half: 0.007866746746003628 + final: 0.007483167573809624 + goal_orientation_error_deg: + mean: 1.9472822835057655 + median_second_half: 1.3664509537256302 + final: 0.7509871966491298 +recorded_at: '2026-10-18T22:45:22' +trajectory_id: trajectory_hera_001515 +t0_sim_ns: 60000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001516.yaml b/metadata/trajectory_hera_001516.yaml new file mode 100644 index 0000000000000000000000000000000000000000..61f6b8ea68a04a90b3cc7e91ab14cf7d73c38815 --- /dev/null +++ b/metadata/trajectory_hera_001516.yaml @@ -0,0 +1,163 @@ +trajectory_dir: zerog_cubesat_arm_wall_docking_00504_20261018T221449_037_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 40000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 24000 + imu: 4800 + joint_states: 14400 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint + arm_ceiling: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint + arm_wall: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: docking + target_kind: cubesat + target_mount: arm_wall + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_cubesat_arm_wall_docking_00504 + overrides: + - env.mode=docking + - env.goal_reached_pos_m=0.25 + - env.target_mount=arm_wall + - env.target_kind=cubesat + - env.target_scale=2.85 + - env.fill_light_intensity=40919 + - env.event_camera_preset=imx636_inspection + - env.docking_speed_range=[0.046,0.071] + seed: 1232003790 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 10815.274193763733 + goal_position_error_m: + mean: 0.1534920825206087 + median_second_half: 0.019895058125257492 + final: 0.01986781507730484 + goal_orientation_error_deg: + mean: 0.7170688473302781 + median_second_half: 0.4008364816313683 + final: 0.40180639517393657 +recorded_at: '2026-10-18T22:26:46' +trajectory_id: trajectory_hera_001516 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001517.yaml b/metadata/trajectory_hera_001517.yaml new file mode 100644 index 0000000000000000000000000000000000000000..16d4192fa87668ee09be93a0a3d787455124e44d --- /dev/null +++ b/metadata/trajectory_hera_001517.yaml @@ -0,0 +1,163 @@ +trajectory_dir: zerog_cubesat_arm_wall_docking_00504_20261018T221449_037_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 40008333333 +end_ns: 80000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 24000 + imu: 4800 + joint_states: 14400 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint + arm_ceiling: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint + arm_wall: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: docking + target_kind: cubesat + target_mount: arm_wall + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_cubesat_arm_wall_docking_00504 + overrides: + - env.mode=docking + - env.goal_reached_pos_m=0.25 + - env.target_mount=arm_wall + - env.target_kind=cubesat + - env.target_scale=2.85 + - env.fill_light_intensity=40919 + - env.event_camera_preset=imx636_inspection + - env.docking_speed_range=[0.046,0.071] + seed: 1232003790 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 10450.269940130413 + goal_position_error_m: + mean: 0.16005316977168518 + median_second_half: 0.020844805985689163 + final: 0.0207622479647398 + goal_orientation_error_deg: + mean: 9.79423608861876 + median_second_half: 0.41604048272381666 + final: 0.41604048272381666 +recorded_at: '2026-10-18T22:38:50' +trajectory_id: trajectory_hera_001517 +t0_sim_ns: 40000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001518.yaml b/metadata/trajectory_hera_001518.yaml new file mode 100644 index 0000000000000000000000000000000000000000..be9c9295181ab9ae32015007bb662a8be2aabfe4 --- /dev/null +++ b/metadata/trajectory_hera_001518.yaml @@ -0,0 +1,163 @@ +trajectory_dir: zerog_cubesat_arm_wall_docking_00504_20261018T221449_037_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 80008333333 +end_ns: 120000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 24000 + imu: 4800 + joint_states: 14400 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint + arm_ceiling: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint + arm_wall: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: docking + target_kind: cubesat + target_mount: arm_wall + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_cubesat_arm_wall_docking_00504 + overrides: + - env.mode=docking + - env.goal_reached_pos_m=0.25 + - env.target_mount=arm_wall + - env.target_kind=cubesat + - env.target_scale=2.85 + - env.fill_light_intensity=40919 + - env.event_camera_preset=imx636_inspection + - env.docking_speed_range=[0.046,0.071] + seed: 1232003790 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 10685.825927250087 + goal_position_error_m: + mean: 0.13879012146864803 + median_second_half: 0.020537119358778 + final: 0.020537119358778 + goal_orientation_error_deg: + mean: 4.264451459425682 + median_second_half: 0.41188772770844895 + final: 0.41188772770844895 +recorded_at: '2026-10-18T22:48:34' +trajectory_id: trajectory_hera_001518 +t0_sim_ns: 80000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001519.yaml b/metadata/trajectory_hera_001519.yaml new file mode 100644 index 0000000000000000000000000000000000000000..839d57f677c3311237f3a0fe00e85f7e0e7be8a8 --- /dev/null +++ b/metadata/trajectory_hera_001519.yaml @@ -0,0 +1,137 @@ +trajectory_dir: orbital_satellite_mockup_black_approach_00507_20261018T224525_040_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 30000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: satellite_mockup + structure_scale: 2.95 + viewpoint_mode: continuous + skydome: false + ambient_light_intensity: 1405.0 +scenario: + name: orbital_satellite_mockup_black_approach_00507 + overrides: + - env.structure_scale=2.95 + - env.structure_radius=1.977 + - env.viewpoint_radius_range=[3.28,3.6] + - env.orbit_speed_range=[0.078,0.141] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=satellite_mockup + - env.spawn_offset_range=[1.0,2.0] + - env.spawn_attitude_noise_deg=60 + - env.skydome=false + - ++env.ambient_light_intensity=1405 + seed: 113251241 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 6490.675455361605 + goal_position_error_m: + mean: 0.32331654705661056 + median_second_half: 0.009316178038716316 + final: 0.009107548743486404 + goal_orientation_error_deg: + mean: 6.9378352276777715 + median_second_half: 0.6671031354921173 + final: 0.5580863260669852 +recorded_at: '2026-10-18T22:51:18' +trajectory_id: trajectory_hera_001519 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001520.yaml b/metadata/trajectory_hera_001520.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b410cb534d14a85e64bf1627f6e4e4059839166f --- /dev/null +++ b/metadata/trajectory_hera_001520.yaml @@ -0,0 +1,137 @@ +trajectory_dir: orbital_satellite_mockup_black_approach_00507_20261018T224525_040_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 30008333333 +end_ns: 60000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: satellite_mockup + structure_scale: 2.95 + viewpoint_mode: continuous + skydome: false + ambient_light_intensity: 1405.0 +scenario: + name: orbital_satellite_mockup_black_approach_00507 + overrides: + - env.structure_scale=2.95 + - env.structure_radius=1.977 + - env.viewpoint_radius_range=[3.28,3.6] + - env.orbit_speed_range=[0.078,0.141] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=satellite_mockup + - env.spawn_offset_range=[1.0,2.0] + - env.spawn_attitude_noise_deg=60 + - env.skydome=false + - ++env.ambient_light_intensity=1405 + seed: 113251241 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 7469.4000505656 + goal_position_error_m: + mean: 0.1538755345681181 + median_second_half: 0.0076271058060228825 + final: 0.007308720611035824 + goal_orientation_error_deg: + mean: 7.7295148524024775 + median_second_half: 2.313433736981155 + final: 0.8627206070949288 +recorded_at: '2026-10-18T22:57:05' +trajectory_id: trajectory_hera_001520 +t0_sim_ns: 30000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001521.yaml b/metadata/trajectory_hera_001521.yaml new file mode 100644 index 0000000000000000000000000000000000000000..62170b5bc6858ba1e3936870a5e9098ffe3bd08e --- /dev/null +++ b/metadata/trajectory_hera_001521.yaml @@ -0,0 +1,137 @@ +trajectory_dir: orbital_satellite_mockup_black_approach_00507_20261018T224525_040_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 60008333333 +end_ns: 90000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: satellite_mockup + structure_scale: 2.95 + viewpoint_mode: continuous + skydome: false + ambient_light_intensity: 1405.0 +scenario: + name: orbital_satellite_mockup_black_approach_00507 + overrides: + - env.structure_scale=2.95 + - env.structure_radius=1.977 + - env.viewpoint_radius_range=[3.28,3.6] + - env.orbit_speed_range=[0.078,0.141] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=satellite_mockup + - env.spawn_offset_range=[1.0,2.0] + - env.spawn_attitude_noise_deg=60 + - env.skydome=false + - ++env.ambient_light_intensity=1405 + seed: 113251241 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 7500.209027707577 + goal_position_error_m: + mean: 0.17159202069112184 + median_second_half: 0.005976030603051186 + final: 0.006416705437004566 + goal_orientation_error_deg: + mean: 4.578029473112714 + median_second_half: 0.17223002343061508 + final: 0.2331665432866623 +recorded_at: '2026-10-18T23:03:50' +trajectory_id: trajectory_hera_001521 +t0_sim_ns: 60000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001522.yaml b/metadata/trajectory_hera_001522.yaml new file mode 100644 index 0000000000000000000000000000000000000000..9cfe00848c06b7faf2d0c4f5ddf9f406cfc48666 --- /dev/null +++ b/metadata/trajectory_hera_001522.yaml @@ -0,0 +1,137 @@ +trajectory_dir: orbital_satellite_mockup_black_orbit_00511_20261018T230605_044_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 30000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: satellite_mockup + structure_scale: 2.51 + viewpoint_mode: continuous + skydome: false + ambient_light_intensity: 743.0 +scenario: + name: orbital_satellite_mockup_black_orbit_00511 + overrides: + - env.structure_scale=2.51 + - env.structure_radius=1.682 + - env.viewpoint_radius_range=[4.1,4.51] + - env.orbit_speed_range=[0.154,0.23] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=satellite_mockup + - env.spawn_offset_range=[0.2,0.71] + - env.spawn_attitude_noise_deg=15 + - env.skydome=false + - ++env.ambient_light_intensity=743 + seed: 1320488399 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8883.186124801636 + goal_position_error_m: + mean: 0.022959852759785133 + median_second_half: 0.006627850234508514 + final: 0.005875390022993088 + goal_orientation_error_deg: + mean: 0.5730869345570222 + median_second_half: 0.35784433614839906 + final: 0.4802164599175805 +recorded_at: '2026-10-18T23:15:58' +trajectory_id: trajectory_hera_001522 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001523.yaml b/metadata/trajectory_hera_001523.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7701ff4ae30bcacd884e83cc2cdc401148e9e215 --- /dev/null +++ b/metadata/trajectory_hera_001523.yaml @@ -0,0 +1,137 @@ +trajectory_dir: orbital_satellite_mockup_black_orbit_00511_20261018T230605_044_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 30008333333 +end_ns: 60000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: satellite_mockup + structure_scale: 2.51 + viewpoint_mode: continuous + skydome: false + ambient_light_intensity: 743.0 +scenario: + name: orbital_satellite_mockup_black_orbit_00511 + overrides: + - env.structure_scale=2.51 + - env.structure_radius=1.682 + - env.viewpoint_radius_range=[4.1,4.51] + - env.orbit_speed_range=[0.154,0.23] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=satellite_mockup + - env.spawn_offset_range=[0.2,0.71] + - env.spawn_attitude_noise_deg=15 + - env.skydome=false + - ++env.ambient_light_intensity=743 + seed: 1320488399 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8659.098496437073 + goal_position_error_m: + mean: 0.0411344458313467 + median_second_half: 0.012761888094246387 + final: 0.011313903145492077 + goal_orientation_error_deg: + mean: 2.401182893836809 + median_second_half: 1.7036325268917125 + final: 0.8479906104587087 +recorded_at: '2026-10-18T23:23:20' +trajectory_id: trajectory_hera_001523 +t0_sim_ns: 30000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001524.yaml b/metadata/trajectory_hera_001524.yaml new file mode 100644 index 0000000000000000000000000000000000000000..767a3b73ec6e90142b2c83f89c89220a30a74ea3 --- /dev/null +++ b/metadata/trajectory_hera_001524.yaml @@ -0,0 +1,137 @@ +trajectory_dir: orbital_satellite_mockup_black_orbit_00511_20261018T230605_044_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 60008333333 +end_ns: 90000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: satellite_mockup + structure_scale: 2.51 + viewpoint_mode: continuous + skydome: false + ambient_light_intensity: 743.0 +scenario: + name: orbital_satellite_mockup_black_orbit_00511 + overrides: + - env.structure_scale=2.51 + - env.structure_radius=1.682 + - env.viewpoint_radius_range=[4.1,4.51] + - env.orbit_speed_range=[0.154,0.23] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=satellite_mockup + - env.spawn_offset_range=[0.2,0.71] + - env.spawn_attitude_noise_deg=15 + - env.skydome=false + - ++env.ambient_light_intensity=743 + seed: 1320488399 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8848.728708267212 + goal_position_error_m: + mean: 0.027671304605653715 + median_second_half: 0.01012569759041071 + final: 0.009746021591126919 + goal_orientation_error_deg: + mean: 0.8404221677728417 + median_second_half: 0.41033020445384805 + final: 0.31593466892717104 +recorded_at: '2026-10-18T23:30:20' +trajectory_id: trajectory_hera_001524 +t0_sim_ns: 60000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001525.yaml b/metadata/trajectory_hera_001525.yaml new file mode 100644 index 0000000000000000000000000000000000000000..eb29f246005dd717116553e1f8cc4eeef572745e --- /dev/null +++ b/metadata/trajectory_hera_001525.yaml @@ -0,0 +1,137 @@ +trajectory_dir: orbital_venus_express_black_orbit_00509_20261018T224840_042_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 30000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: venus_express + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: false + ambient_light_intensity: 1403.0 +scenario: + name: orbital_venus_express_black_orbit_00509 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=4.4 + - env.viewpoint_radius_range=[10.78,11.86] + - env.orbit_speed_range=[0.204,0.268] + - env.camera_clipping_range=[0.05,80.0] + - env.lidar_max_distance=80.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=venus_express + - env.spawn_offset_range=[0.2,1.38] + - env.spawn_attitude_noise_deg=15 + - env.skydome=false + - ++env.ambient_light_intensity=1403 + seed: 95107925 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 7843.198243379593 + goal_position_error_m: + mean: 0.16527794075640928 + median_second_half: 0.003048973623663187 + final: 0.0027668094262480736 + goal_orientation_error_deg: + mean: 1.7278503485579393 + median_second_half: 0.5190229800998013 + final: 0.5238859680671446 +recorded_at: '2026-10-18T22:54:39' +trajectory_id: trajectory_hera_001525 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001526.yaml b/metadata/trajectory_hera_001526.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7b0e81fec767121c986e817dd522a6048f4d17db --- /dev/null +++ b/metadata/trajectory_hera_001526.yaml @@ -0,0 +1,137 @@ +trajectory_dir: orbital_venus_express_black_orbit_00509_20261018T224840_042_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 30008333333 +end_ns: 60000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: venus_express + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: false + ambient_light_intensity: 1403.0 +scenario: + name: orbital_venus_express_black_orbit_00509 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=4.4 + - env.viewpoint_radius_range=[10.78,11.86] + - env.orbit_speed_range=[0.204,0.268] + - env.camera_clipping_range=[0.05,80.0] + - env.lidar_max_distance=80.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=venus_express + - env.spawn_offset_range=[0.2,1.38] + - env.spawn_attitude_noise_deg=15 + - env.skydome=false + - ++env.ambient_light_intensity=1403 + seed: 95107925 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8747.375400543213 + goal_position_error_m: + mean: 0.03973978656062553 + median_second_half: 0.004834062419831753 + final: 0.0047699627466499805 + goal_orientation_error_deg: + mean: 0.7054623689616966 + median_second_half: 0.5101985891355564 + final: 0.454706130254406 +recorded_at: '2026-10-18T22:59:47' +trajectory_id: trajectory_hera_001526 +t0_sim_ns: 30000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001527.yaml b/metadata/trajectory_hera_001527.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f1200ad4a2edbec24e6f0f97ed111e7ec86da7c0 --- /dev/null +++ b/metadata/trajectory_hera_001527.yaml @@ -0,0 +1,137 @@ +trajectory_dir: orbital_venus_express_black_orbit_00509_20261018T224840_042_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 60008333333 +end_ns: 90000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: venus_express + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: false + ambient_light_intensity: 1403.0 +scenario: + name: orbital_venus_express_black_orbit_00509 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=4.4 + - env.viewpoint_radius_range=[10.78,11.86] + - env.orbit_speed_range=[0.204,0.268] + - env.camera_clipping_range=[0.05,80.0] + - env.lidar_max_distance=80.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=venus_express + - env.spawn_offset_range=[0.2,1.38] + - env.spawn_attitude_noise_deg=15 + - env.skydome=false + - ++env.ambient_light_intensity=1403 + seed: 95107925 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8552.22559171915 + goal_position_error_m: + mean: 0.0643200801833969 + median_second_half: 0.0050203148275613785 + final: 0.005178479943424463 + goal_orientation_error_deg: + mean: 0.9577533778381875 + median_second_half: 0.5278770930591064 + final: 0.5158088958476342 +recorded_at: '2026-10-18T23:06:00' +trajectory_id: trajectory_hera_001527 +t0_sim_ns: 60000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001528.yaml b/metadata/trajectory_hera_001528.yaml new file mode 100644 index 0000000000000000000000000000000000000000..6ea1eff1c55fabb287a2c9070a52f516a4c320ae --- /dev/null +++ b/metadata/trajectory_hera_001528.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_venus_express_earth_approach_00508_20261018T224827_041_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 30000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: venus_express + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_venus_express_earth_approach_00508 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=4.4 + - env.viewpoint_radius_range=[10.65,11.71] + - env.orbit_speed_range=[0.143,0.267] + - env.camera_clipping_range=[0.05,80.0] + - env.lidar_max_distance=80.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=venus_express + - env.spawn_offset_range=[1.6,3.19] + - env.spawn_attitude_noise_deg=60 + seed: 1194777364 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 6477.194438911974 + goal_position_error_m: + mean: 0.32465705286558294 + median_second_half: 0.0025000888854265213 + final: 0.0023298992309719324 + goal_orientation_error_deg: + mean: 10.102413417714708 + median_second_half: 0.19767569308029925 + final: 0.21127121130363608 +recorded_at: '2026-10-18T22:54:28' +trajectory_id: trajectory_hera_001528 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001529.yaml b/metadata/trajectory_hera_001529.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7b3e9af3fbecc1953c74341a0aa1cad787d59a4d --- /dev/null +++ b/metadata/trajectory_hera_001529.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_venus_express_earth_approach_00508_20261018T224827_041_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 30008333333 +end_ns: 60000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: venus_express + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_venus_express_earth_approach_00508 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=4.4 + - env.viewpoint_radius_range=[10.65,11.71] + - env.orbit_speed_range=[0.143,0.267] + - env.camera_clipping_range=[0.05,80.0] + - env.lidar_max_distance=80.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=venus_express + - env.spawn_offset_range=[1.6,3.19] + - env.spawn_attitude_noise_deg=60 + seed: 1194777364 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 7024.142719089985 + goal_position_error_m: + mean: 0.2765804132202213 + median_second_half: 0.0053106797859072685 + final: 0.005278047639876604 + goal_orientation_error_deg: + mean: 6.521285688888046 + median_second_half: 0.8190659866233212 + final: 0.7630906653843627 +recorded_at: '2026-10-18T22:59:15' +trajectory_id: trajectory_hera_001529 +t0_sim_ns: 30000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001530.yaml b/metadata/trajectory_hera_001530.yaml new file mode 100644 index 0000000000000000000000000000000000000000..06effb82b558813d6a0a45cf068be89edd91fe49 --- /dev/null +++ b/metadata/trajectory_hera_001530.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_venus_express_earth_approach_00508_20261018T224827_041_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 60008333333 +end_ns: 90000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: venus_express + structure_scale: 1.0 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_venus_express_earth_approach_00508 + overrides: + - env.structure_scale=1.0 + - env.structure_radius=4.4 + - env.viewpoint_radius_range=[10.65,11.71] + - env.orbit_speed_range=[0.143,0.267] + - env.camera_clipping_range=[0.05,80.0] + - env.lidar_max_distance=80.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=venus_express + - env.spawn_offset_range=[1.6,3.19] + - env.spawn_attitude_noise_deg=60 + seed: 1194777364 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 5190.318402837962 + goal_position_error_m: + mean: 0.5019338134099169 + median_second_half: 0.002600894309580326 + final: 0.0027458511758595705 + goal_orientation_error_deg: + mean: 14.65719727688804 + median_second_half: 0.44645171324031335 + final: 0.4790342502217865 +recorded_at: '2026-10-18T23:06:05' +trajectory_id: trajectory_hera_001530 +t0_sim_ns: 60000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001531.yaml b/metadata/trajectory_hera_001531.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a0fb915ec1104514ea6c7dec11e0a3d423bcf4b4 --- /dev/null +++ b/metadata/trajectory_hera_001531.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_satellite_mockup_earth_approach_00513_20261018T233023_046_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 30000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: satellite_mockup + structure_scale: 2.75 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_satellite_mockup_earth_approach_00513 + overrides: + - env.structure_scale=2.75 + - env.structure_radius=1.843 + - env.viewpoint_radius_range=[4.63,5.09] + - env.orbit_speed_range=[0.065,0.191] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=satellite_mockup + - env.spawn_offset_range=[1.0,2.0] + - env.spawn_attitude_noise_deg=60 + seed: 614828294 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 8440.722136497498 + goal_position_error_m: + mean: 0.07865826377917415 + median_second_half: 0.009744095616042614 + final: 0.009876122698187828 + goal_orientation_error_deg: + mean: 1.4105935935003289 + median_second_half: 0.6138108512973574 + final: 0.6628146838278485 +recorded_at: '2026-10-18T23:38:08' +trajectory_id: trajectory_hera_001531 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001532.yaml b/metadata/trajectory_hera_001532.yaml new file mode 100644 index 0000000000000000000000000000000000000000..adfb273e8f84fa8a59314975b24da34a6fb5b3db --- /dev/null +++ b/metadata/trajectory_hera_001532.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_satellite_mockup_earth_approach_00513_20261018T233023_046_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 30008333333 +end_ns: 60000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: satellite_mockup + structure_scale: 2.75 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_satellite_mockup_earth_approach_00513 + overrides: + - env.structure_scale=2.75 + - env.structure_radius=1.843 + - env.viewpoint_radius_range=[4.63,5.09] + - env.orbit_speed_range=[0.065,0.191] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=satellite_mockup + - env.spawn_offset_range=[1.0,2.0] + - env.spawn_attitude_noise_deg=60 + seed: 614828294 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 7775.364199578762 + goal_position_error_m: + mean: 0.1476681731591787 + median_second_half: 0.006817192304879427 + final: 0.006838453002274036 + goal_orientation_error_deg: + mean: 4.076816974391533 + median_second_half: 0.6581758114426463 + final: 0.6986536468982574 +recorded_at: '2026-10-18T23:45:28' +trajectory_id: trajectory_hera_001532 +t0_sim_ns: 30000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001533.yaml b/metadata/trajectory_hera_001533.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a7f507384bb9fce4ddc05f9cdf76336111d1d466 --- /dev/null +++ b/metadata/trajectory_hera_001533.yaml @@ -0,0 +1,135 @@ +trajectory_dir: orbital_satellite_mockup_earth_approach_00513_20261018T233023_046_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.mobile.inspection.task +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 60008333333 +end_ns: 90000000000 +duration_s: 29.991666667 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 600 + depth: 600 + semantic: 600 + lidar_scans: 300 + poses: 14400 + imu: 3600 + joint_states: 0 + steps: 301 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: {} +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.mobile.inspection.task + robot: SatelliteMockup + action_terms: + robot/thrust: 12 + agent_rate_s: 0.1 + episode_length_s: 30.0 + scenario_type: orbital_inspection + structure_kind: satellite_mockup + structure_scale: 2.75 + viewpoint_mode: continuous + skydome: low_res + ambient_light_intensity: 0.0 +scenario: + name: orbital_satellite_mockup_earth_approach_00513 + overrides: + - env.structure_scale=2.75 + - env.structure_radius=1.843 + - env.viewpoint_radius_range=[4.63,5.09] + - env.orbit_speed_range=[0.065,0.191] + - env.camera_clipping_range=[0.05,60.0] + - env.lidar_max_distance=60.0 + - env.event_camera_preset=imx636_inspection + - env.structure_kind=satellite_mockup + - env.spawn_offset_range=[1.0,2.0] + - env.spawn_attitude_noise_deg=60 + seed: 614828294 + policy: latest rsl_rl_ppo +summary: + steps: 301 + return: 7656.825897142291 + goal_position_error_m: + mean: 0.15183579236070976 + median_second_half: 0.0081167072057724 + final: 0.007973791100084782 + goal_orientation_error_deg: + mean: 5.617409760285018 + median_second_half: 0.6615940543179731 + final: 0.6217668477416586 +recorded_at: '2026-10-18T23:52:43' +trajectory_id: trajectory_hera_001533 +t0_sim_ns: 60000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 12 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - orbit_radius + - orbit_speed + - orbit_axis + - viewpoint_direction diff --git a/metadata/trajectory_hera_001534.yaml b/metadata/trajectory_hera_001534.yaml new file mode 100644 index 0000000000000000000000000000000000000000..505cecb118d5292cdff21a6065e1a6527ee3428c --- /dev/null +++ b/metadata/trajectory_hera_001534.yaml @@ -0,0 +1,163 @@ +trajectory_dir: zerog_satellite_mockup_arm_ceiling_inspection_00510_20261018T230357_043_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 40000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 24000 + imu: 4800 + joint_states: 14400 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint + arm_ceiling: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint + arm_wall: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: inspection + target_kind: satellite_mockup + target_mount: arm_ceiling + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_satellite_mockup_arm_ceiling_inspection_00510 + overrides: + - env.mode=inspection + - env.goal_reached_pos_m=0.25 + - env.target_mount=arm_ceiling + - env.target_kind=satellite_mockup + - env.target_scale=1.07 + - env.fill_light_intensity=35751 + - env.event_camera_preset=imx636_inspection + - env.arc_speed_range=[0.044,0.051] + seed: 778971276 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 10338.906487584114 + goal_position_error_m: + mean: 0.13110566911452504 + median_second_half: 0.024874761700630188 + final: 0.05643389746546745 + goal_orientation_error_deg: + mean: 14.778021724751996 + median_second_half: 0.6424475933313993 + final: 1.0918740407182383 +recorded_at: '2026-10-18T23:17:36' +trajectory_id: trajectory_hera_001534 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001535.yaml b/metadata/trajectory_hera_001535.yaml new file mode 100644 index 0000000000000000000000000000000000000000..d3e3682d0f6f31669eaa5e6a46582fb4592ac29e --- /dev/null +++ b/metadata/trajectory_hera_001535.yaml @@ -0,0 +1,163 @@ +trajectory_dir: zerog_satellite_mockup_arm_ceiling_inspection_00510_20261018T230357_043_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 40008333333 +end_ns: 80000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 24000 + imu: 4800 + joint_states: 14400 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint + arm_ceiling: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint + arm_wall: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: inspection + target_kind: satellite_mockup + target_mount: arm_ceiling + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_satellite_mockup_arm_ceiling_inspection_00510 + overrides: + - env.mode=inspection + - env.goal_reached_pos_m=0.25 + - env.target_mount=arm_ceiling + - env.target_kind=satellite_mockup + - env.target_scale=1.07 + - env.fill_light_intensity=35751 + - env.event_camera_preset=imx636_inspection + - env.arc_speed_range=[0.044,0.051] + seed: 778971276 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 10449.214332342148 + goal_position_error_m: + mean: 0.1685574285853236 + median_second_half: 0.022861991077661514 + final: 0.03081042878329754 + goal_orientation_error_deg: + mean: 5.429633144199237 + median_second_half: 0.9315899914655491 + final: 0.7720025681473484 +recorded_at: '2026-10-18T23:30:08' +trajectory_id: trajectory_hera_001535 +t0_sim_ns: 40000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001536.yaml b/metadata/trajectory_hera_001536.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8f4638c47c0eb05a4de3765d4ba7aa68eefa6d54 --- /dev/null +++ b/metadata/trajectory_hera_001536.yaml @@ -0,0 +1,163 @@ +trajectory_dir: zerog_satellite_mockup_arm_ceiling_inspection_00510_20261018T230357_043_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 80008333333 +end_ns: 120000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 24000 + imu: 4800 + joint_states: 14400 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint + arm_ceiling: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint + arm_wall: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: inspection + target_kind: satellite_mockup + target_mount: arm_ceiling + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_satellite_mockup_arm_ceiling_inspection_00510 + overrides: + - env.mode=inspection + - env.goal_reached_pos_m=0.25 + - env.target_mount=arm_ceiling + - env.target_kind=satellite_mockup + - env.target_scale=1.07 + - env.fill_light_intensity=35751 + - env.event_camera_preset=imx636_inspection + - env.arc_speed_range=[0.044,0.051] + seed: 778971276 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 10878.284863054752 + goal_position_error_m: + mean: 0.10642192492544875 + median_second_half: 0.02037728577852249 + final: 0.02249421551823616 + goal_orientation_error_deg: + mean: 5.412351384475352 + median_second_half: 0.8605901750432666 + final: 0.8189396814845289 +recorded_at: '2026-10-18T23:43:39' +trajectory_id: trajectory_hera_001536 +t0_sim_ns: 80000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001537.yaml b/metadata/trajectory_hera_001537.yaml new file mode 100644 index 0000000000000000000000000000000000000000..94f66a8bfbb253e7b51f048262eb9ae36f681de8 --- /dev/null +++ b/metadata/trajectory_hera_001537.yaml @@ -0,0 +1,163 @@ +trajectory_dir: zerog_satellite_mockup_arm_ceiling_inspection_00512_20261018T230615_045_env000_ep0002 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 2 +end_reason: reset +start_ns: 8333333 +end_ns: 40000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 24000 + imu: 4800 + joint_states: 14400 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint + arm_ceiling: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint + arm_wall: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: inspection + target_kind: satellite_mockup + target_mount: arm_ceiling + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_satellite_mockup_arm_ceiling_inspection_00512 + overrides: + - env.mode=inspection + - env.goal_reached_pos_m=0.25 + - env.target_mount=arm_ceiling + - env.target_kind=satellite_mockup + - env.target_scale=1.15 + - env.fill_light_intensity=36473 + - env.event_camera_preset=imx636_inspection + - env.arc_speed_range=[0.08,0.096] + seed: 750018058 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 10834.814984560013 + goal_position_error_m: + mean: 0.12831634555262816 + median_second_half: 0.018482543528079987 + final: 0.03096572682261467 + goal_orientation_error_deg: + mean: 2.041871726650027 + median_second_half: 1.3741658658365712 + final: 1.338525832025739 +recorded_at: '2026-10-18T23:21:06' +trajectory_id: trajectory_hera_001537 +t0_sim_ns: 0 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001538.yaml b/metadata/trajectory_hera_001538.yaml new file mode 100644 index 0000000000000000000000000000000000000000..9cdded0c2c1e69fa3170d351e3a7baa07164d840 --- /dev/null +++ b/metadata/trajectory_hera_001538.yaml @@ -0,0 +1,163 @@ +trajectory_dir: zerog_satellite_mockup_arm_ceiling_inspection_00512_20261018T230615_045_env000_ep0003 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 3 +end_reason: reset +start_ns: 40008333333 +end_ns: 80000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 24000 + imu: 4800 + joint_states: 14400 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint + arm_ceiling: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint + arm_wall: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: inspection + target_kind: satellite_mockup + target_mount: arm_ceiling + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_satellite_mockup_arm_ceiling_inspection_00512 + overrides: + - env.mode=inspection + - env.goal_reached_pos_m=0.25 + - env.target_mount=arm_ceiling + - env.target_kind=satellite_mockup + - env.target_scale=1.15 + - env.fill_light_intensity=36473 + - env.event_camera_preset=imx636_inspection + - env.arc_speed_range=[0.08,0.096] + seed: 750018058 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 10273.569416821003 + goal_position_error_m: + mean: 0.21092948407873027 + median_second_half: 0.013744105584919453 + final: 0.0051766810938715935 + goal_orientation_error_deg: + mean: 1.8791734358689707 + median_second_half: 1.7186667337502628 + final: 1.6803220517498112 +recorded_at: '2026-10-18T23:32:28' +trajectory_id: trajectory_hera_001538 +t0_sim_ns: 40000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance diff --git a/metadata/trajectory_hera_001539.yaml b/metadata/trajectory_hera_001539.yaml new file mode 100644 index 0000000000000000000000000000000000000000..02e2d1a413ecfd88f3d49fb3686d7422be81cc62 --- /dev/null +++ b/metadata/trajectory_hera_001539.yaml @@ -0,0 +1,163 @@ +trajectory_dir: zerog_satellite_mockup_arm_ceiling_inspection_00512_20261018T230615_045_env000_ep0004 +source: simulation +simulator: Space Robotics Bench (Isaac Sim 6.1, Isaac Lab 3.0) +task: srb.tasks.digital_twin.zerog_lab.task_inspection +env_index: 0 +episode_index: 4 +end_reason: reset +start_ns: 80008333333 +end_ns: 120000000000 +duration_s: 39.991666667000004 +clock: simulation time (sim.current_time) in ns; shared by all streams and events +rates_hz: + physics_and_render: 120.0 + agent: 10.0 + camera: 20.0 + lidar: 10.0 + state: 120.0 +camera_update_period_s: 0.008333333333333333 +counts: + rgb: 800 + depth: 800 + semantic: 800 + lidar_scans: 400 + poses: 24000 + imu: 4800 + joint_states: 14400 + steps: 401 +semantic_classes: + BACKGROUND: 0 + UNLABELLED: 1 + target: 2 + robot_arm: 3 + spacecraft: 4 +articulations: + robot: + - x_lock_joint + - y_lock_joint + - base_joint + - rw_revolute_joint + arm_ceiling: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint + arm_wall: + - rail_joint + - shoulder_pan_joint + - shoulder_lift_joint + - elbow_joint + - wrist_1_joint + - wrist_2_joint + - wrist_3_joint +frames_and_units: + world: z-up world frame of the environment origin; metres, seconds + quaternion: xyzw + poses: 'body pose and twist in world frame (v: m/s, w: rad/s)' + imu: 'robot base IMU: angular_velocity (rad/s, body), linear_acceleration (m/s^2, + body); orientation = base quat (world)' + depth: uint16 PNG along the optical axis in units of calibration depth.scale_m (0.001 + = mm), 0 = no return + semantic: uint8 PNG of class IDs (see semantic_classes) + action: raw policy action as applied (see metadata.task_info.action_space) +task_info: + task_cfg: srb.tasks.digital_twin.zerog_lab.task_inspection + robot: FloatingPlatform + action_terms: + robot/thrust: 8 + agent_rate_s: 0.1 + episode_length_s: 40.0 + scenario_type: zerog_lab_inspection + mode: inspection + target_kind: satellite_mockup + target_mount: arm_ceiling + standoff_range_m: + - 1.1 + - 1.6 + arena_x_m: + - 0.4 + - 4.7 + arena_y_m: + - -2.6 + - -0.4 +scenario: + name: zerog_satellite_mockup_arm_ceiling_inspection_00512 + overrides: + - env.mode=inspection + - env.goal_reached_pos_m=0.25 + - env.target_mount=arm_ceiling + - env.target_kind=satellite_mockup + - env.target_scale=1.15 + - env.fill_light_intensity=36473 + - env.event_camera_preset=imx636_inspection + - env.arc_speed_range=[0.08,0.096] + seed: 750018058 + policy: latest rsl_rl_ppo +summary: + steps: 401 + return: 9908.413468778133 + goal_position_error_m: + mean: 0.23675280872864785 + median_second_half: 0.005431736819446087 + final: 0.17692860960960388 + goal_orientation_error_deg: + mean: 3.5449167042788488 + median_second_half: 1.6501054015833805 + final: 0.4010686813827223 +recorded_at: '2026-10-18T23:46:05' +trajectory_id: trajectory_hera_001539 +t0_sim_ns: 80000000000 +steps_layout: + observation: + - name: act_previous + group: state + size: 8 + - name: tf_pos_robot_to_goal + group: state + size: 3 + - name: tf_rot6d_robot + group: state + size: 6 + - name: tf_rot6d_robot_to_goal + group: state + size: 6 + - name: vel_ang_goal + group: state + size: 3 + - name: vel_ang_robot + group: state + size: 3 + - name: vel_lin_goal + group: state + size: 3 + - name: vel_lin_robot + group: state + size: 3 + - name: imu_ang_vel + group: proprio + size: 3 + - name: imu_lin_acc + group: proprio + size: 3 + reward_terms: + - penalty_action_magnitude + - penalty_action_rate + - penalty_angular_velocity + - penalty_orientation_error + - penalty_position_distance + - penalty_speed_near_target + - penalty_velocity_at_target + - reward_orientation_precision + - reward_position_approach + - reward_position_fine + - reward_position_precision + - reward_smooth_station_keeping + task_state: + - viewpoints_reached + - phase + - docking + - docking_goal_distance + - target_distance