Dynamic Manipulation on DOMINO 35 suites (full)
26.6Success Rate (SR)Qwen-VLA-Instruct
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Qwen-VLA-InstructEvaluation protocol=Zero-shot to dynamic manipulation2026.05 | 26.6 | 39.5 | |
| LingBot-VAEvaluation protocol=Zero-shot to dynamic manipulation2026.05 | 24.1 | 36.1 | |
| Qwen-VLA-BaseEvaluation protocol=Zero-shot to dynamic manipulation2026.05 | 21.1 | 37.4 | |
| PUMAEvaluation protocol=Fine-tuned on dynamic manipulation data2026.05 | 17.2 | 35 | |
| LingBot-VLAEvaluation protocol=Zero-shot to dynamic manipulation, Input modality=w/ depth2026.05 | 11.8 | 26.7 | |
| StarVLA-OFTEvaluation protocol=Fine-tuned on dynamic manipulation data2026.05 | 10.9 | 30.5 | |
| π0.5Evaluation protocol=Fine-tuned on dynamic manipulation data2026.05 | 9.6 | 26.2 | |
| OpenVLA-OFTEvaluation protocol=Fine-tuned on dynamic manipulation data2026.05 | 9.1 | 24.1 | |
| π0Evaluation protocol=Fine-tuned on dynamic manipulation data2026.05 | 8.2 | 24 | |
| π0.5Evaluation protocol=Zero-shot to dynamic manipulation2026.05 | 7.5 | 20.4 | |
| OpenVLA-OFTEvaluation protocol=Zero-shot to dynamic manipulation2026.05 | 6.7 | 20 | |
| InternVLA-M1Evaluation protocol=Fine-tuned on dynamic manipulation data2026.05 | 5.4 | 27.6 | |
| RDT-1BEvaluation protocol=Fine-tuned on dynamic manipulation data2026.05 | 5.3 | 17.7 | |
| VLA-AdapterEvaluation protocol=Fine-tuned on dynamic manipulation data2026.05 | 4.4 | 24.3 | |
| π0-FASTEvaluation protocol=Fine-tuned on dynamic manipulation data2026.05 | 3.5 | 20.9 | |
| OpenVLAEvaluation protocol=Fine-tuned on dynamic manipulation data2026.05 | 1.5 | 6.1 |