Open-Vocabulary Long-Horizon Manipulation on RoboVoLo Common Sense Suite
52.38Infer RateVoLoAgent (Only VLA)
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| VoLoAgent (Only VLA)Method Family=VoLoAgent, Orchestrator Type=VoLoAgent, Episodes=32026.06 | 52.38 | 33.33 | 45.83 | 47.62 | 44.44 | |
| VoLoAgent (Full)Method Family=VoLoAgent, Orchestrator Type=VoLoAgent, Episodes=32026.06 | 52.38 | 50 | 62.5 | 52.38 | 54.44 | |
| DreamZeroMethod Family=Single action model, Orchestrator Type=None, Episodes=32026.06 | 19.05 | 12.5 | 20.83 | 0 | 13.33 | |
| VoLoAgent (No VLA)Method Family=VoLoAgent, Orchestrator Type=VoLoAgent, Episodes=32026.06 | 19.05 | 41.67 | 62.5 | 0 | 32.22 | |
| MolmoBotMethod Family=Single action model, Orchestrator Type=None, Episodes=32026.06 | 14.29 | 0 | 12.5 | 0 | 6.67 | |
| CaPX-eMethod Family=Code-as-policy + VLM, Variant=CAP-X ensemble, Orchestrator Type=Code-as-policy, Episodes=32026.06 | 14.29 | 16.67 | 29.17 | 0 | 15.56 | |
| π0-FASTMethod Family=Single action model, Orchestrator Type=None, Episodes=32026.06 | 9.52 | 4.17 | 0 | 0 | 3.33 | |
| CaPX-sMethod Family=Code-as-policy + VLM, Variant=CaP-X single, Orchestrator Type=Code-as-policy, Episodes=32026.06 | 9.52 | 12.5 | 37.5 | 0 | 15.56 | |
| TiPToPMethod Family=TAMP + VLM, Orchestrator Type=TAMP, Episodes=32026.06 | 4.76 | 8.33 | 0 | 0 | 3.33 | |
| π0.5Method Family=Single action model, Orchestrator Type=None, Episodes=32026.06 | 0 | 16.67 | 4.17 | 23.81 | 11.11 | |
| MolmoAct2Method Family=Single action model, Orchestrator Type=None, Episodes=32026.06 | 0 | 0 | 12.5 | 0 | 3.33 |