Path-level reasoning on UNOBench synthetic No obstructions (test)
94.8SR (%)UNOGrasp
Evaluation results
| Method | Links | ||
|---|---|---|---|
| UNOGraspPrompting Type=Oracle (with SoM)2025.11 | 94.8 | 0.03 | |
| UNOGraspPrompting Type=Natural Language Prompting2025.11 | 92.5 | 0.06 | |
| Qwen2.5-VLPrompting Type=Natural Language Prompting, Evaluation Protocol=SFT2025.11 | 91.4 | — | |
| Qwen2.5-VLPrompting Type=Oracle (with SoM), Evaluation Protocol=SFT2025.11 | 88.7 | — | |
| Gemini Robotics-ER 1.5Prompting Type=Oracle (with SoM)2025.11 | 68.7 | 0.17 | |
| Gemini Robotics-ER 1.5Prompting Type=Oracle (with SoM), Evaluation Protocol=ICL2025.11 | 54.6 | 0.25 | |
| Gemini Robotics-ER 1.5Prompting Type=Natural Language Prompting2025.11 | 50.2 | 0.88 | |
| Gemini Robotics-ER 1.5Prompting Type=Natural Language Prompting, Evaluation Protocol=ICL2025.11 | 45.3 | 0.83 | |
| Qwen2.5-VLPrompting Type=Natural Language Prompting, Evaluation Protocol=ICL2025.11 | 11.2 | 0.84 | |
| Qwen2.5-VLPrompting Type=Oracle (with SoM), Evaluation Protocol=ICL2025.11 | 9.8 | 0.64 |