Surgical Debridement on Debridement subtask
90Task Success RateClean
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| CleanPolicy Architecture=ACT2026.06 | 90 | — | — | — | — | |
| CleanPolicy Architecture=Diffusion Policy2026.06 | 90 | — | — | — | — | |
| RandomPolicy Architecture=ACT, Attack Objective=Disruptive2026.06 | 90 | 0.1 | 97 | 98.9 | — | |
| RandomPolicy Architecture=Diffusion Policy, Attack Objective=Disruptive2026.06 | 80 | 0.1 | 97 | 98.9 | — | |
| UAPPolicy Architecture=ACT, Attack Objective=Steering2026.06 | 80 | 0.1 | 96 | 98.9 | 0.2 | |
| CleanPolicy Architecture=π02026.06 | 70 | — | — | — | — | |
| RandomPolicy Architecture=π0, Attack Objective=Disruptive2026.06 | 60 | 0.1 | 96 | 99.2 | — | |
| UAPPolicy Architecture=ACT, Attack Objective=Disruptive2026.06 | 60 | 0.1 | 95 | 99 | — | |
| UAPPolicy Architecture=Diffusion Policy, Attack Objective=Disruptive2026.06 | 50 | 0.1 | 95 | 98.9 | — | |
| UAPPolicy Architecture=Diffusion Policy, Attack Objective=Steering2026.06 | 50 | 0.1 | 94 | 99 | 0.7 | |
| UAPPolicy Architecture=π0, Attack Objective=Steering2026.06 | 30 | 0.1 | 94 | 99.3 | 0.7 | |
| PGDPolicy Architecture=ACT, Attack Objective=Steering2026.06 | 20 | 87.2 | 97 | 96.1 | 0.8 | |
| TPAPolicy Architecture=ACT, Attack Objective=Steering2026.06 | 10 | 3.6 | 90 | 98.8 | 4.6 | |
| UAPPolicy Architecture=π0, Attack Objective=Disruptive2026.06 | 0 | 0.1 | 94 | 99.3 | — | |
| PGDPolicy Architecture=ACT, Attack Objective=Disruptive2026.06 | 0 | 84.3 | 97 | 95.1 | — | |
| PGDPolicy Architecture=Diffusion Policy, Attack Objective=Disruptive2026.06 | 0 | 4,642 | 96 | 93.6 | — | |
| PGDPolicy Architecture=π0, Attack Objective=Disruptive2026.06 | 0 | 3,050 | 97 | 99.1 | — | |
| TPAPolicy Architecture=ACT, Attack Objective=Disruptive2026.06 | 0 | 3.9 | 81 | 98.7 | — | |
| TPAPolicy Architecture=Diffusion Policy, Attack Objective=Disruptive2026.06 | 0 | 7.4 | 68 | 98.6 | — | |
| TPAPolicy Architecture=π0, Attack Objective=Disruptive2026.06 | 0 | 7.8 | 76 | 98.7 | — | |
| PGDPolicy Architecture=Diffusion Policy, Attack Objective=Steering2026.06 | 0 | 5,024 | 96 | 96.3 | -1.1 | |
| PGDPolicy Architecture=π0, Attack Objective=Steering2026.06 | 0 | 3,314 | 97 | 93.8 | -0.6 | |
| TPAPolicy Architecture=Diffusion Policy, Attack Objective=Steering2026.06 | 0 | 13.5 | 83 | 98.6 | 1.9 | |
| TPAPolicy Architecture=π0, Attack Objective=Steering2026.06 | 0 | 6.6 | 88 | 98.8 | 3.1 |