Simulation Object Manipulation on SimplerEnv WidowX + Bridge setup
58.3Placement Success Rate (Spoon on Towel)SOFAR
Evaluation results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| SOFARTraining Data=Zero-Shot2025.02 | 58.3 | 62.5 | 75 | 66.7 | 91.7 | 70.8 | 66.7 | 37.5 | 58.3 | — | — | |
| UniVLAAdaptation strategy=Full LoRA Fine-tuning2025.05 | 52.8 | — | 79.2 | 55.6 | — | 2.8 | 93 | 80.6 | 47.9 | 76.4 | 66.7 | |
| Octo-SmallTraining Data=OXE2025.02 | 47.2 | 77.8 | 27.8 | 9.7 | 40.3 | 4.2 | 87.5 | 56.9 | 30 | — | — | |
| Octo-Small2025.05 | 47.2 | — | 27.8 | 9.7 | — | 4.2 | 87.5 | 56.9 | 30 | 77.8 | 40.3 | |
| UniVLAAdaptation strategy=Decoder-only2025.05 | 37.5 | — | 62.5 | 33.3 | — | 4.2 | 87.5 | 66.7 | 35.4 | 70.8 | 37.5 | |
| RoboVLMTraining Data=Bridge2025.02 | 29.2 | 54.2 | 25 | 25 | 45.8 | 12.5 | 58.3 | 58.3 | 31.3 | — | — | |
| RoboVLM2025.05 | 29.2 | — | 25 | 25 | — | 12.5 | 58.3 | 58.3 | 31.3 | 54.2 | 45.8 | |
| RoboVLMTraining Data=OXE2025.02 | 20.8 | 37.5 | 33.3 | 25 | 8.3 | 8.3 | 0 | 0 | 13.5 | — | — | |
| SpatialVLATraining Data=OXE2025.02 | 20.8 | 25 | 41.7 | 20.8 | 58.3 | 25 | 79.2 | 70.8 | 34.4 | — | — | |
| SpatialVLATraining Data=Bridge2025.02 | 16.7 | 20.8 | 29.2 | 25 | 62.5 | 29.2 | 100 | 100 | 42.7 | — | — | |
| Octo-BaseTraining Data=OXE2025.02 | 12.5 | 34.7 | 52.8 | 8.3 | 31.9 | 0 | 66.7 | 43.1 | 16 | — | — | |
| Octo-Base2025.05 | 12.5 | — | 52.8 | 8.3 | — | 0 | 66.7 | 43.1 | 16 | 34.7 | 31.9 | |
| RT-1-XTraining Data=OXE2025.02 | 0 | 16.7 | 20.8 | 4.2 | 8.3 | 0 | 0 | 0 | 1.1 | — | — | |
| OpenVLATraining Data=OXE2025.02 | 0 | 4.1 | 33.3 | 0 | 12.5 | 0 | 8.3 | 4.1 | 1 | — | — | |
| RT-1-X2025.05 | 0 | — | 20.8 | 4.2 | — | 0 | 0 | 0 | 1.1 | 16.7 | 8.3 | |
| OpenVLA2025.05 | 0 | — | 33.3 | 0 | — | 0 | 8.3 | 4.1 | 1 | 4.1 | 12.5 |