Table Cleaning

Pick three debris blocks into a dustpan, then tip all three into a trash bin.

Multi-object graspingTool useDump

Stage 01 / 02

Long environmentTableClean-v1

EmbodimentRight Sharpa

Observation91 state dims

Action29 targets

Horizon1200 steps

Stage definitions

Exact slices of the parent sequence.

Canonical instructions, handedness, action width, horizon, and success conditions are drawn from the current task registry and success-contract documentation.

#	Stage / environment	Hand	Action	Max steps	Success summary
01	Collect debris`TableCleanPickDebris-v1`	Right	29	900	Place all three blocks inside the dustpan, release them, and hold stable for eight steps.
02	Dump dustpan`TableCleanDumpDustpan-v1`	Right	29	600	Move all three blocks into the bin-bottom region and hold for ten steps.

Paper results

Table Cleaning in the benchmark.

Stage success in the training scene (%, 50 episodes). Best scores in each row are highlighted.

Stage benchmark: success rates in percent
Stage	PPO	RLPD	ACT	Diffusion Policy	GR00T	π0.5
PickDebris	0.0	0.0	0.0	2.0	0.0	0.0
DumpDustpan	0.0	0.0	92.0	70.0	82.0	78.0

Cross-scene results · mean ± scene standard deviation

Cross-scene: success rates in percent
Stage	PPO	RLPD	ACT	Diffusion Policy	GR00T	π0.5
PickDebris	1.3 ± 2.3	0.0 ± 0.0	0.0 ± 0.0	0.0 ± 0.0	1.3 ± 1.2	0.0 ± 0.0
DumpDustpan	0.7 ± 1.2	0.0 ± 0.0	59.3 ± 17.9	1.3 ± 2.3	88.7 ± 2.3	68.0 ± 4.0

Complete sequence · end-to-end success (%)

Long-horizon: success rates in percent
Task	PPO	RLPD	ACT	Diffusion Policy	GR00T	π0.5
Table cleaning	0.0	0.0	0.0	0.0	0.0	0.0

Source: manuscript experiment tables. Compare all methods and tasks →

Shared contract

Same workstation.
Same predicates.

Stage alignment

Each stage-level environment uses the matching physical success predicate from the long task.

Persistence

Hold counters or latches prevent unstable contact from being counted as task completion.

Replay

State and action traces can be restored and replayed for validation and offline rendering.