Instructions to use sislaboratory/owm-rl-baselines with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- stable-baselines3
How to use sislaboratory/owm-rl-baselines with stable-baselines3:
from huggingface_sb3 import load_from_hub checkpoint = load_from_hub( repo_id="sislaboratory/owm-rl-baselines", filename="{MODEL FILENAME}.zip", ) - Notebooks
- Google Colab
- Kaggle
Upload RL run owm-iss-numerical-v1-noncoop-terminal-ppo-vector
Browse files- rl/best/owm-iss-numerical-v1-noncoop-terminal-ppo-vector/config.yaml +174 -0
- rl/best/owm-iss-numerical-v1-noncoop-terminal-ppo-vector/env_config.yaml +153 -0
- rl/best/owm-iss-numerical-v1-noncoop-terminal-ppo-vector/final_model.zip +3 -0
- rl/best/owm-iss-numerical-v1-noncoop-terminal-ppo-vector/promotion.yaml +24 -0
- rl/best/owm-iss-numerical-v1-noncoop-terminal-ppo-vector/vecnormalize.pkl +3 -0
rl/best/owm-iss-numerical-v1-noncoop-terminal-ppo-vector/config.yaml
ADDED
|
@@ -0,0 +1,174 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
environments:
|
| 2 |
+
env_name: iss-numerical
|
| 3 |
+
dt: 1.0
|
| 4 |
+
max_steps: 360
|
| 5 |
+
max_range_m: 750.0
|
| 6 |
+
render_view: DRAGON_FPV
|
| 7 |
+
physics:
|
| 8 |
+
mass: 12000.0
|
| 9 |
+
inertia_diag:
|
| 10 |
+
- 80000.0
|
| 11 |
+
- 80000.0
|
| 12 |
+
- 50000.0
|
| 13 |
+
linear_damping: 0.0
|
| 14 |
+
angular_damping: 0.0
|
| 15 |
+
dragon_collision_radius_m: 2.25
|
| 16 |
+
collision_boxes_path: collision_boxes.yaml
|
| 17 |
+
control:
|
| 18 |
+
limit_force_n: 1600.0
|
| 19 |
+
limit_torque_nm: 2000.0
|
| 20 |
+
dock:
|
| 21 |
+
position:
|
| 22 |
+
- 0.225
|
| 23 |
+
- -24.5
|
| 24 |
+
- -2.5
|
| 25 |
+
quaternion:
|
| 26 |
+
- 0.7071068
|
| 27 |
+
- -0.7071068
|
| 28 |
+
- 0.0
|
| 29 |
+
- 0.0
|
| 30 |
+
enabled: true
|
| 31 |
+
max_distance_m: 0.1
|
| 32 |
+
max_velocity_m_s: 0.5
|
| 33 |
+
max_attitude_error_deg: 5.0
|
| 34 |
+
max_body_rate_rad_s: 0.008727
|
| 35 |
+
ports:
|
| 36 |
+
- harmony_fwd_pma2
|
| 37 |
+
- harmony_nadir_cbm
|
| 38 |
+
- zvezda_aft
|
| 39 |
+
- pirs_nadir
|
| 40 |
+
- rassvet_nadir
|
| 41 |
+
orbit:
|
| 42 |
+
epoch: '2026-08-01T00:00:00Z'
|
| 43 |
+
epoch_offset_range_s:
|
| 44 |
+
- 0.0
|
| 45 |
+
- 5400.0
|
| 46 |
+
start_radius_range_m:
|
| 47 |
+
- 100.0
|
| 48 |
+
- 125.0
|
| 49 |
+
start_speed_max_m_s: 0.0
|
| 50 |
+
start_attitude_error_max_deg: 0.0
|
| 51 |
+
start_rate_max_rad_s: 0.0
|
| 52 |
+
perturbations:
|
| 53 |
+
zonal_max_degree: 4
|
| 54 |
+
third_body_sun: true
|
| 55 |
+
third_body_moon: true
|
| 56 |
+
drag: true
|
| 57 |
+
sensor_noise:
|
| 58 |
+
enabled: true
|
| 59 |
+
sigma_pos_m: 0.0
|
| 60 |
+
sigma_pos_frac_of_range: 0.01
|
| 61 |
+
sigma_vel_m_s: 0.03
|
| 62 |
+
sigma_att_rad: 5.0e-05
|
| 63 |
+
sigma_rate_rad_s: 1.0e-05
|
| 64 |
+
observation:
|
| 65 |
+
goal_error: true
|
| 66 |
+
mode: relative
|
| 67 |
+
reward_weights:
|
| 68 |
+
position: -0.9
|
| 69 |
+
velocity: -0.05
|
| 70 |
+
attitude: -0.2
|
| 71 |
+
body_rate: -0.02
|
| 72 |
+
collision: -2.0
|
| 73 |
+
dock_success: 1000.0
|
| 74 |
+
escape: -1000.0
|
| 75 |
+
progress: -2.0
|
| 76 |
+
proximity: 0.5
|
| 77 |
+
alignment: 0.5
|
| 78 |
+
effort: -0.05
|
| 79 |
+
reward_shaping:
|
| 80 |
+
position_delta_m: 1.0
|
| 81 |
+
position_scale_m: 225.0
|
| 82 |
+
velocity_delta_m_s: 0.1
|
| 83 |
+
velocity_scale_m_s: 5.0
|
| 84 |
+
attitude_delta_rad: 0.05
|
| 85 |
+
attitude_scale_rad: 3.141592653589793
|
| 86 |
+
rate_delta_rad_s: 0.005
|
| 87 |
+
rate_scale_rad_s: 0.05
|
| 88 |
+
rotation_gate_far: 0.5
|
| 89 |
+
rotation_gate_range_m: 25.0
|
| 90 |
+
proximity_scale_m: 1.0
|
| 91 |
+
alignment_attitude_scale_rad: 0.1
|
| 92 |
+
collision_terminates: false
|
| 93 |
+
rl:
|
| 94 |
+
algo: ppo
|
| 95 |
+
n_envs: 32
|
| 96 |
+
vec: subproc
|
| 97 |
+
total_timesteps: 25000000
|
| 98 |
+
device: cpu
|
| 99 |
+
obs: vector
|
| 100 |
+
obs_resnet:
|
| 101 |
+
variant: resnet18
|
| 102 |
+
layer: avgpool
|
| 103 |
+
image_size: 224
|
| 104 |
+
device: null
|
| 105 |
+
action_repeat: 1
|
| 106 |
+
norm_reward: false
|
| 107 |
+
checkpoint:
|
| 108 |
+
save_freq: 2500000
|
| 109 |
+
hyperparams:
|
| 110 |
+
n_steps: 1024
|
| 111 |
+
batch_size: 1024
|
| 112 |
+
learning_rate: 0.00021578468144928557
|
| 113 |
+
gamma: 0.997
|
| 114 |
+
gae_lambda: 0.9453223308741135
|
| 115 |
+
ent_coef: 0.001
|
| 116 |
+
clip_range: 0.1
|
| 117 |
+
n_epochs: 20
|
| 118 |
+
target_kl: 0.02
|
| 119 |
+
vf_coef: 0.6666451279183626
|
| 120 |
+
policy_kwargs:
|
| 121 |
+
net_arch:
|
| 122 |
+
- 256
|
| 123 |
+
- 256
|
| 124 |
+
- 256
|
| 125 |
+
logging:
|
| 126 |
+
entity: duncan-world-models
|
| 127 |
+
project: amos-out-of-this-world-model
|
| 128 |
+
mode: online
|
| 129 |
+
tags: []
|
| 130 |
+
run_name: ppo_25M_near_noncoop_terminal
|
| 131 |
+
run_dir: runs/ppo_25M_near_noncoop_terminal
|
| 132 |
+
resume: false
|
| 133 |
+
seed: 0
|
| 134 |
+
extend_timesteps: null
|
| 135 |
+
external_wandb: false
|
| 136 |
+
val:
|
| 137 |
+
enabled: true
|
| 138 |
+
every_steps: 50000
|
| 139 |
+
episodes: 3
|
| 140 |
+
video_episodes: 1
|
| 141 |
+
max_frames: 1200
|
| 142 |
+
hub:
|
| 143 |
+
upload: true
|
| 144 |
+
repo_id: sislaboratory/owm-rl-baselines
|
| 145 |
+
eval:
|
| 146 |
+
checkpoint: null
|
| 147 |
+
episodes: 20
|
| 148 |
+
deterministic: true
|
| 149 |
+
video_path: null
|
| 150 |
+
allow_unnormalized: false
|
| 151 |
+
eval_matrix:
|
| 152 |
+
checkpoint: null
|
| 153 |
+
run_dir: null
|
| 154 |
+
ports: all
|
| 155 |
+
trials: 50
|
| 156 |
+
seed: 100000
|
| 157 |
+
rate_hz: null
|
| 158 |
+
action_repeat: 1
|
| 159 |
+
criteria:
|
| 160 |
+
- position
|
| 161 |
+
- position_velocity
|
| 162 |
+
- full
|
| 163 |
+
tolerances_m:
|
| 164 |
+
- 0.1
|
| 165 |
+
- 0.2
|
| 166 |
+
- 0.5
|
| 167 |
+
- 1.0
|
| 168 |
+
- 2.0
|
| 169 |
+
- 5.0
|
| 170 |
+
- 10.0
|
| 171 |
+
n_envs: 10
|
| 172 |
+
vec: subproc
|
| 173 |
+
out_dir: null
|
| 174 |
+
allow_unnormalized: false
|
rl/best/owm-iss-numerical-v1-noncoop-terminal-ppo-vector/env_config.yaml
ADDED
|
@@ -0,0 +1,153 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
dt: 1.0
|
| 2 |
+
max_steps: 360
|
| 3 |
+
max_range_m: 750.0
|
| 4 |
+
collision_terminates: false
|
| 5 |
+
physics:
|
| 6 |
+
mass: 12000.0
|
| 7 |
+
inertia_diag:
|
| 8 |
+
- 80000.0
|
| 9 |
+
- 80000.0
|
| 10 |
+
- 50000.0
|
| 11 |
+
linear_damping: 0.0
|
| 12 |
+
angular_damping: 0.0
|
| 13 |
+
dragon_collision_radius_m: 2.25
|
| 14 |
+
start_radius_range_m:
|
| 15 |
+
- 100.0
|
| 16 |
+
- 500.0
|
| 17 |
+
collision_boxes_path: collision_boxes.yaml
|
| 18 |
+
control:
|
| 19 |
+
limit_force_n: 1600.0
|
| 20 |
+
limit_torque_nm: 2000.0
|
| 21 |
+
dock:
|
| 22 |
+
position:
|
| 23 |
+
- 0.225
|
| 24 |
+
- -24.5
|
| 25 |
+
- -2.5
|
| 26 |
+
quaternion:
|
| 27 |
+
- 0.7071068
|
| 28 |
+
- -0.7071068
|
| 29 |
+
- 0.0
|
| 30 |
+
- 0.0
|
| 31 |
+
enabled: true
|
| 32 |
+
max_distance_m: 0.1
|
| 33 |
+
max_velocity_m_s: 0.5
|
| 34 |
+
max_attitude_error_deg: 5.0
|
| 35 |
+
max_body_rate_rad_s: 0.008727
|
| 36 |
+
ports:
|
| 37 |
+
- name: harmony_fwd_pma2
|
| 38 |
+
position:
|
| 39 |
+
- 0.194
|
| 40 |
+
- -24.5
|
| 41 |
+
- -3.336
|
| 42 |
+
quaternion:
|
| 43 |
+
- 0.7071067690849304
|
| 44 |
+
- -0.7071067690849304
|
| 45 |
+
- -0.0
|
| 46 |
+
- 0.0
|
| 47 |
+
- name: harmony_nadir_cbm
|
| 48 |
+
position:
|
| 49 |
+
- 0.233
|
| 50 |
+
- -14.939
|
| 51 |
+
- -9.985
|
| 52 |
+
quaternion:
|
| 53 |
+
- -0.0
|
| 54 |
+
- -0.0
|
| 55 |
+
- 0.0
|
| 56 |
+
- 1.0
|
| 57 |
+
- name: zvezda_aft
|
| 58 |
+
position:
|
| 59 |
+
- 0.233
|
| 60 |
+
- 38.55
|
| 61 |
+
- -1.87
|
| 62 |
+
quaternion:
|
| 63 |
+
- -0.0
|
| 64 |
+
- 0.0
|
| 65 |
+
- 0.7071067690849304
|
| 66 |
+
- -0.7071067690849304
|
| 67 |
+
- name: pirs_nadir
|
| 68 |
+
position:
|
| 69 |
+
- 0.233
|
| 70 |
+
- 20.129
|
| 71 |
+
- -12.79
|
| 72 |
+
quaternion:
|
| 73 |
+
- -0.0
|
| 74 |
+
- -0.0
|
| 75 |
+
- 0.0
|
| 76 |
+
- 1.0
|
| 77 |
+
- name: rassvet_nadir
|
| 78 |
+
position:
|
| 79 |
+
- 0.24
|
| 80 |
+
- 7.161
|
| 81 |
+
- -14.767
|
| 82 |
+
quaternion:
|
| 83 |
+
- -0.0
|
| 84 |
+
- -0.0
|
| 85 |
+
- 0.0
|
| 86 |
+
- 1.0
|
| 87 |
+
sensor_noise:
|
| 88 |
+
enabled: true
|
| 89 |
+
sigma_pos_m: 0.0
|
| 90 |
+
sigma_pos_frac_of_range: 0.01
|
| 91 |
+
sigma_vel_m_s: 0.03
|
| 92 |
+
sigma_att_rad: 5.0e-05
|
| 93 |
+
sigma_rate_rad_s: 1.0e-05
|
| 94 |
+
observation:
|
| 95 |
+
goal_error: true
|
| 96 |
+
mode: relative
|
| 97 |
+
reward_weights:
|
| 98 |
+
position: -0.9
|
| 99 |
+
velocity: -0.05
|
| 100 |
+
attitude: -0.2
|
| 101 |
+
body_rate: -0.02
|
| 102 |
+
collision: -2.0
|
| 103 |
+
dock_success: 1000.0
|
| 104 |
+
escape: -1000.0
|
| 105 |
+
progress: -2.0
|
| 106 |
+
proximity: 0.5
|
| 107 |
+
alignment: 0.5
|
| 108 |
+
effort: -0.05
|
| 109 |
+
reward_shaping:
|
| 110 |
+
position_delta_m: 1.0
|
| 111 |
+
position_scale_m: 225.0
|
| 112 |
+
velocity_delta_m_s: 0.1
|
| 113 |
+
velocity_scale_m_s: 5.0
|
| 114 |
+
attitude_delta_rad: 0.05
|
| 115 |
+
attitude_scale_rad: 3.141592653589793
|
| 116 |
+
rate_delta_rad_s: 0.005
|
| 117 |
+
rate_scale_rad_s: 0.05
|
| 118 |
+
rotation_gate_far: 0.5
|
| 119 |
+
rotation_gate_range_m: 25.0
|
| 120 |
+
proximity_scale_m: 1.0
|
| 121 |
+
alignment_attitude_scale_rad: 0.1
|
| 122 |
+
reward_goal_position: null
|
| 123 |
+
render_view: DRAGON_FPV
|
| 124 |
+
render: null
|
| 125 |
+
orbit:
|
| 126 |
+
epoch: '2026-08-01T00:00:00Z'
|
| 127 |
+
sma_m: 6795000.0
|
| 128 |
+
ecc: 0.0005
|
| 129 |
+
inc_deg: 51.64
|
| 130 |
+
raan_deg: 0.0
|
| 131 |
+
argp_deg: 0.0
|
| 132 |
+
mean_anomaly_deg: 0.0
|
| 133 |
+
epoch_offset_range_s:
|
| 134 |
+
- 0.0
|
| 135 |
+
- 5400.0
|
| 136 |
+
start_radius_range_m:
|
| 137 |
+
- 100.0
|
| 138 |
+
- 125.0
|
| 139 |
+
start_speed_max_m_s: 0.0
|
| 140 |
+
start_attitude_error_max_deg: 0.0
|
| 141 |
+
start_rate_max_rad_s: 0.0
|
| 142 |
+
perturbations:
|
| 143 |
+
zonal_max_degree: 4
|
| 144 |
+
third_body_sun: true
|
| 145 |
+
third_body_moon: true
|
| 146 |
+
drag: true
|
| 147 |
+
chief_mass_kg: 419700.0
|
| 148 |
+
chief_ballistic:
|
| 149 |
+
area_m2: 1500.0
|
| 150 |
+
cd: 2.2
|
| 151 |
+
chaser_ballistic:
|
| 152 |
+
area_m2: 50.0
|
| 153 |
+
cd: 2.2
|
rl/best/owm-iss-numerical-v1-noncoop-terminal-ppo-vector/final_model.zip
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:118d5c010833acb70424cfc05d1ae98de493598ec5537515e35d41c987eb8163
|
| 3 |
+
size 3400551
|
rl/best/owm-iss-numerical-v1-noncoop-terminal-ppo-vector/promotion.yaml
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
name: owm-iss-numerical-v1-noncoop-terminal-ppo-vector
|
| 2 |
+
policy:
|
| 3 |
+
algo: ppo
|
| 4 |
+
obs: vector
|
| 5 |
+
env: iss-numerical
|
| 6 |
+
dt_s: 1.0
|
| 7 |
+
train_ports:
|
| 8 |
+
- harmony_fwd_pma2
|
| 9 |
+
- harmony_nadir_cbm
|
| 10 |
+
- zvezda_aft
|
| 11 |
+
- pirs_nadir
|
| 12 |
+
- rassvet_nadir
|
| 13 |
+
timesteps: 25000000
|
| 14 |
+
run: ppo_25M_near_noncoop_terminal
|
| 15 |
+
source: runs/ppo_25M_near_noncoop_terminal/checkpoints/model_7500000_steps.zip
|
| 16 |
+
step: 7500000
|
| 17 |
+
from_finals: false
|
| 18 |
+
criterion: val_return
|
| 19 |
+
window: 1250000
|
| 20 |
+
scores:
|
| 21 |
+
val_return: 58.715167907177786
|
| 22 |
+
train_return: -65.44462457142859
|
| 23 |
+
min_pos: 4.567768662717165
|
| 24 |
+
wandb: https://wandb.ai/duncan-world-models/amos-out-of-this-world-model/runs/s31jpzmf
|
rl/best/owm-iss-numerical-v1-noncoop-terminal-ppo-vector/vecnormalize.pkl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:203824cf6378ff9271a56a5fc637fb89a80e0db85691d4b933ae46477000fa22
|
| 3 |
+
size 6380
|