Robot Manipulation Adaptation on LIBERO novel viewpoints
79.1Average Success RateDART
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| DARTBase Policy=π0.5, Adaptation=One-shot, Trials=32026.07 | 79.1 | 92 | 80.8 | 64.4 | |
| DARTPolicy=π0.52026.07 | 79.1 | — | — | — | |
| DARTpolicy=π0.52026.07 | 79.1 | 92 | 80.8 | 64.4 | |
| TIESpolicy=π0.52026.07 | 77.6 | 91.9 | 79.8 | 61 | |
| DART + Merging (TSV)Merging Method=TSV, Policy=π0.52026.07 | 75.7 | — | — | — | |
| RESMpolicy=π0.52026.07 | 75.5 | 90.8 | 78 | 57.7 | |
| DART + Merging (Iso-C)Merging Method=Iso-C, Policy=π0.52026.07 | 74.8 | — | — | — | |
| DART + Merging (TA)Merging Method=TA, Policy=π0.52026.07 | 74.5 | — | — | — | |
| Iso-Cpolicy=π0.52026.07 | 74.5 | 91.8 | 76.4 | 55.3 | |
| FLABase Policy=π0.5, Adaptation=One-shot, Trials=32026.07 | 74.3 | 92.2 | 76.4 | 54.3 | |
| DART + Merging (TIES)Merging Method=TIES, Policy=π0.52026.07 | 70.9 | — | — | — | |
| RETAINBase Policy=π0.5, Adaptation=One-shot, Trials=32026.07 | 69.6 | 87.4 | 72.4 | 48.9 | |
| Zero-shotBase Policy=π0.5, Adaptation=None, Trials=32026.07 | 54.5 | 88.3 | 63.9 | 11.3 | |
| One-shot FTBase Policy=π0.5, Adaptation=One-shot, Trials=32026.07 | 31.5 | 43.4 | 33.3 | 17.8 |