USC-PSI-Lab/SIMPLE-decoupled-wbc-Teleop

SIMPLE decoupled-WBC, teleop tasks

The 12 teleop tasks of USC-PSI-Lab/SIMPLE-decoupled-wbc (all tasks except the 5 motion-planning *MP tasks), as one LeRobot v2.1 dataset. Recorded rows and videos are byte-identical to the merged dataset; only episode, frame and task indices were renumbered and the statistics recomputed (see Subset notes). The MP tasks are in USC-PSI-Lab/SIMPLE-decoupled-wbc-MP.

• Robot: Unitree G1, 50 fps, egocentric 360x640 H.264 video (libx264 crf 23, yuv420p, keyframe every 10 frames, no B-frames)
• Episodes: 1128, frames: 465228, tasks: 12
• states (32-d) / action (36-d): layout in meta/modality.json

State / action layout

All joint values are in radians. For i < 31, states[i] and action[i] are the same joint. The names are also in meta/info.json under features.states.names and features.action.names.

dims	action (36-d)	states (32-d)
0-6	left hand joint targets: thumb_0, thumb_1, thumb_2, middle_0, middle_1, index_0, index_1	measured left hand joints, same order
7-13	right hand joint targets: thumb_0, thumb_1, thumb_2, index_0, index_1, middle_0, middle_1	measured right hand joints, same order
14-20	left arm joint targets: shoulder pitch/roll/yaw, elbow, wrist roll/pitch/yaw	measured left arm joints
21-27	right arm joint targets, same order	measured right arm joints
28-30	waist (torso) roll, pitch, yaw targets	measured waist roll, pitch, yaw (see below)
31	base height command (m, 0.41-0.75)	previous step's base height command
32	base forward velocity command vx (m/s, ±0.5)	–
33	base lateral velocity command vy (m/s, ±0.5)	–
34	base yaw-rate command vyaw (rad/s, ±1)	–
35	target heading yaw (rad, ±π)	–

Dims 0-31 are absolute targets. Dims 32-34 are velocity commands, marked absolute: false in meta/modality.json.

The two hands use different finger orders: left is thumb, middle, index and right is thumb, index, middle. Both the training data and the SIMPLE eval agents use this order. The per-joint column observation.hand_joints uses thumb, index, middle for both hands.

Teleop (decoupled WBC) conventions

All tasks here use the decoupled whole-body controller; the eval agent is simple/baselines/psi0_decoupled_wbc.py, which builds the model input the same way.

• states[28:31] = the measured waist joint positions at the current frame.
• states[31] = the last base height command, 0.74 at frame 0.
• action[34] = the base yaw-rate command vyaw, over the full ±1 range.

These differ from the MP tasks, where states[28:31] is the last commanded waist value and action[34] is a 0/1 turning flag. When training on both datasets together, treat action[34] as one turning channel.

Data quality

• Joint order is consistent across all tasks (including the left hand) and matches the eval agents.
• Time axis: states[t] is the observation at frame t and action[t] is the command sent at frame t.
• Other per-joint columns: observation.arm_joints (14), observation.leg_joints (12 leg joints, then waist yaw, roll, pitch), observation.prev_torso_rpy and observation.prev_height.

Trailing padding

Episodes whose plain replay (every recorded frame, --no-stop-on-success, success judged at the last step) failed got the last frame held for 50 extra rows (base velocity 0), with the recorded rows unchanged. 554 episodes are padded; the other 574 are not.

Padding is capped at 50 rows per episode. The padding was first appended in up to 4 rounds of 50 (0-200 rows); 271 of these episodes had more than 50, and their extra rows (35900 in total) were removed from the end of the episode and its video (per-episode record in cap_padding_log.jsonl, renumbered to this dataset's episode and task indices). A held run at the end of an episode can be longer than 50 rows when the source recording already ends with a hold (up to 21 rows in BendPickAndPlaceTeleop).

Subset notes

• Made from the merged dataset with scripts/subset_simple_decoupled_wbc.py --exclude-regex 'MP-v0$' (SIMPLE), so its padding cap and video re-encode carry over unchanged.
• Tasks keep their alphabetical order and are renumbered 0-11; episodes keep their order and are renumbered contiguously; index is rewritten to a global 0..N-1.
• meta/stats.json (and the identical meta/stats_psi0.json) is recomputed over the 465228 teleop frames (mean/std/min/max/q01/q99, float64 accumulation), so it differs from the merged dataset's statistics. Every row of meta/episodes_stats.jsonl is computed from its full episode.
• Videos use a keyframe every 10 frames and no B-frames (-g 10 -keyint_min 10 -sc_threshold 0 -bf 0), so a random-frame lookup decodes ~5 frames.

Tasks

task_index	source	instruction	episodes	frames
0	G1WholebodyBendHandoverTeleop-v0	bend the robot and pick up the cracker box, then place it on the container.	96	44575
1	G1WholebodyBendPickAndPlaceTeleop-v0	bend to grasp the cracker box and drop it in the basket.	48	23958
2	G1WholebodyBendPickTeleop-v0	bend the robot and pick up the cracker box	97	22473
3	G1WholebodyCloseDoorTeleop-v0	move forward to the door and close it	99	32251
4	G1WholebodyHandoverTeleop-v0	Hand over cracker box from right hand to left hand and place it on the container.	100	50530
5	G1WholebodyLocomotionPickBetweenTablesTeleop-v0	pick up the cracker box from table1,locomotion to table2,and place on table2.	99	69464
6	G1WholebodyOpenFaucetTeleop-v0	turn the faucet	99	38458
7	G1WholebodyOpenOvenTeleop-v0	move forward to the oven and open it	99	38313
8	G1WholebodyOpenTrashCanTeleop-v0	move forward to the trash can and open it	99	36985
9	G1WholebodyPickAndPlaceAndHugContainerTeleop-v0	pick up the apple from table1,hug the container,walk to table2,and place on table2.	94	46588
10	G1WholebodyPushOfficeChairTeleop-v0	move forward to the office chair and push it to the table	99	35979
11	G1WholebodyXMovePickTeleop-v0	move forward to pick up the cracker box	99	25654