can_clean_final

Whole-body teleoperation on a Unitree G1, recorded 2026-08-25. The robot picks a can off a low table and places it on a white table.

This is can_to_martino_2 and can_to_martino_3 combined, with every episode reviewed by hand and the failures removed. Prefer this over either source.

episodes	105
frames	212,290
fps	50
codebase version	v3.0
task	Bring the can to the white table

Schema

feature	dtype	shape	contents
observation.state	float32	[31]	29 body joints in G1_29_JointIndex order, then left/right gripper
action	float32	[66]	64-D SONIC motion token, then left/right gripper
observation.images.ego_view	video	[480, 640, 3]	head camera
observation.images.left_wrist	video	[480, 640, 3]	left wrist camera
observation.images.right_wrist	video	[480, 640, 3]	right wrist camera

action[t] is the command that produces observation.state[t+1], so each episode is one frame shorter than its recording.

Review

All 124 episodes of the two source datasets were watched and marked pass/fail with a 1-10 quality rating. 19 failed and were removed:

cause	episodes
failed to pick up the can	6
dropped the can	5
camera blocked the view or fell	5
never attempted the pick	2
took over a minute	1

Of the 105 that remain, 82 carry a rating: mean 7.2, with 3 rated 4 or below. The camera failures are a rig problem rather than a teleoperation one — the head camera physically obstructed the grasp or fell mid-episode — so they cluster in time rather than being spread evenly.

Known issue: the left gripper is dead

observation.state[29] and action[64] are identically 0.0 in every frame — the episodes were effectively recorded one-handed. Their q01/q99 in meta/stats.json are set by hand to 0/1 so quantile normalisation does not divide by zero. A policy trained on this will never open or close the left hand. The right gripper is healthy, closed in 41% of frames. No other channel is degenerate: all 64 token dimensions and all 29 joints vary.