Robotic Manipulation on Dataset B force condition 1.0 (Firm)
87Success Rate (SR)IMG+FS
Evaluation results
| Method | Links | ||
|---|---|---|---|
| IMG+FSTactile Modality=Tactile image input, Force Supervision=true, LoRA Fine-tuning=true2026.05 | 87 | 30.6 | |
| FIELD+FS (Tabero-VTLA)Tactile Modality=Force field input, Force Supervision=true, LoRA Fine-tuning=true2026.05 | 86 | 32.4 | |
| FORCE E+FSTactile Modality=Force input via MLP encoder, Force Supervision=true, LoRA Fine-tuning=true2026.05 | 84 | 30.3 | |
| FSTactile Modality=No tactile input, Force Supervision=true, LoRA Fine-tuning=true2026.05 | 82 | 30.4 | |
| FORCE D+FSTactile Modality=Force input via decoder, Force Supervision=true, LoRA Fine-tuning=true2026.05 | 82 | 28.5 | |
| FIELDTactile Modality=Force field input, Force Supervision=false, LoRA Fine-tuning=true2026.05 | 40 | 2.9 | |
| FORCE ETactile Modality=Force input via MLP encoder, Force Supervision=false, LoRA Fine-tuning=true2026.05 | 40 | 2.5 | |
| IMGTactile Modality=Tactile image input, Force Supervision=false, LoRA Fine-tuning=true2026.05 | 37 | 3 | |
| NONETactile Modality=No tactile input, Force Supervision=false, LoRA Fine-tuning=true2026.05 | 0 | 0 |