Failure Detection on Franka seen
0.15Brier ScoreSAFE-RNN-TDQC (Ours)
Evaluation Results
| Method | Links | |
|---|---|---|
| SAFE-RNN-TDQC (Ours)VLA Model=π0-FAST2026.04 | 0.15 | |
| RNN-TDQC (Ours)VLA Model=π0-FAST2026.04 | 0.204 | |
| SAFE-MLP BCEVLA Model=π0-FAST2026.04 | 0.206 | |
| SAFE-MLP-TDQC (Ours)VLA Model=π0-FAST2026.04 | 0.21 | |
| SAFE-RNNVLA Model=π0-FAST2026.04 | 0.22 | |
| RNN-BCEVLA Model=π0-FAST2026.04 | 0.237 | |
| Avg entropyVLA Model=π0-FAST2026.04 | 0.281 | |
| Avg prob.VLA Model=π0-FAST2026.04 | 0.29 | |
| Max prob.VLA Model=π0-FAST2026.04 | 0.331 | |
| Running Avg entropyVLA Model=π0-FAST2026.04 | 0.341 | |
| Running Avg prob.VLA Model=π0-FAST2026.04 | 0.359 |