Failure Detection on LIBERO unseen
0.097Brier ScoreSAFE-RNN-TDQC (Ours)
Evaluation Results
| Method | Links | |
|---|---|---|
| SAFE-RNN-TDQC (Ours)VLA Model=π02026.04 | 0.097 | |
| SAFE-RNN-TDQC (Ours)VLA Model=UniVLA2026.04 | 0.1 | |
| RNN-TDQC (Ours)VLA Model=UniVLA2026.04 | 0.107 | |
| SAFE-MLP-TDQC (Ours)VLA Model=π02026.04 | 0.128 | |
| SAFE-MLP-TDQC (Ours)VLA Model=UniVLA2026.04 | 0.131 | |
| SAFE-MLP BCEVLA Model=π02026.04 | 0.137 | |
| RNN-TDQC (Ours)VLA Model=π0-FAST2026.04 | 0.141 | |
| SAFE-RNNVLA Model=π0-FAST2026.04 | 0.148 | |
| SAFE-MLP-TDQC (Ours)VLA Model=π0-FAST2026.04 | 0.15 | |
| RNN-BCEVLA Model=UniVLA2026.04 | 0.152 | |
| SAFE-MLP BCEVLA Model=UniVLA2026.04 | 0.158 | |
| RNN-BCEVLA Model=π0-FAST2026.04 | 0.158 | |
| SAFE-RNNVLA Model=UniVLA2026.04 | 0.162 | |
| SAFE-MLP BCEVLA Model=π0-FAST2026.04 | 0.162 | |
| SAFE-RNN-TDQC (Ours)VLA Model=π0-FAST2026.04 | 0.163 | |
| SAFE-RNNVLA Model=π02026.04 | 0.172 | |
| RNN-TDQC (Ours)VLA Model=OpenVLA2026.04 | 0.197 | |
| RNN-BCEVLA Model=OpenVLA2026.04 | 0.206 | |
| SAFE-RNN-TDQC (Ours)VLA Model=OpenVLA2026.04 | 0.218 | |
| Avg prob.VLA Model=π0-FAST2026.04 | 0.218 | |
| Avg entropyVLA Model=π0-FAST2026.04 | 0.222 | |
| SAFE-MLP-TDQC (Ours)VLA Model=OpenVLA2026.04 | 0.229 | |
| SAFE-MLP BCEVLA Model=OpenVLA2026.04 | 0.231 | |
| Running Avg prob.VLA Model=π0-FAST2026.04 | 0.238 | |
| SAFE-RNNVLA Model=OpenVLA2026.04 | 0.255 | |
| Running Avg entropyVLA Model=π0-FAST2026.04 | 0.264 | |
| Running Avg entropyVLA Model=OpenVLA2026.04 | 0.273 | |
| Avg entropyVLA Model=OpenVLA2026.04 | 0.313 | |
| Max prob.VLA Model=π0-FAST2026.04 | 0.318 | |
| Running Avg entropyVLA Model=UniVLA2026.04 | 0.33 | |
| Running Avg prob.VLA Model=OpenVLA2026.04 | 0.356 | |
| Avg prob.VLA Model=OpenVLA2026.04 | 0.364 | |
| Max prob.VLA Model=OpenVLA2026.04 | 0.39 | |
| Avg entropyVLA Model=UniVLA2026.04 | 0.391 | |
| Avg prob.VLA Model=UniVLA2026.04 | 0.532 | |
| Running Avg prob.VLA Model=UniVLA2026.04 | 0.544 | |
| Max prob.VLA Model=UniVLA2026.04 | 0.899 |