Robotic Manipulation on Real World (unseen environments and tasks)
90Task 1 Success RateVIEW-LLaVA-13B
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| VIEW-LLaVA-13BVLM=LLaVA-13B, Data=RoVI Book2025.05 | 90 | 90 | 100 | 100 | 70 | 70 | 90 | 90 | 87.5 | |
| VIEW-GPT4oVLM=GPT-4o2025.05 | 80 | 100 | 90 | 90 | 60 | 70 | 90 | 80 | 82.5 | |
| CoPaConditioning=language, Visual Prompting=Set-of-Mark (SoM)2025.05 | 40 | 90 | 80 | 60 | 0 | 40 | 20 | 30 | 45 | |
| VoxposerConditioning=language2025.05 | 30 | 80 | 80 | 10 | 0 | 30 | 30 | 20 | 43.8 |