Whiteboard Erasing on Real-world Robotic Manipulation Dataset
60Success RateTaF-VLA
Evaluation Results
| Method | Links | |
|---|---|---|
| TaF-VLABackbone=VLA(π)2026.01 | 60 | |
| ACT + TaF-AdapterBackbone=ACT2026.01 | 33.3 | |
| DP + TaF-AdapterBackbone=Diffusion2026.01 | 33.3 | |
| π0.5Backbone=VLA(π)2026.01 | 33.3 | |
| FreeTacManBackbone=ACT2026.01 | 26.7 | |
| DPBackbone=Diffusion2026.01 | 20 | |
| ACTBackbone=ACT2026.01 | 13.3 |