Failure Detection on LIBERO-10 Seen Tasks
88.5bACCHide-and-Seek
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Hide-and-SeekPolicy=π0, Category=Class.2026.05 | 88.5 | 92.6 | 69.3 | |
| SAFE-MLPPolicy=π0, Category=Class., Venue=NeurIPS'252026.05 | 87 | 91.5 | 68.8 | |
| Hide-and-SeekPolicy=OpenVLA, Category=Class.2026.05 | 85.2 | 85.3 | 66 | |
| SAFE-MLPPolicy=OpenVLA, Category=Class., Venue=NeurIPS'252026.05 | 82.3 | 81.9 | 63 | |
| MahalanobisPolicy=π0, Category=OOD Det., Venue=ICML'182026.05 | 78.8 | 84 | 71.5 | |
| LogpZOPolicy=π0, Category=OOD Det., Venue=RSS'252026.05 | 78.6 | 83.5 | 69.1 | |
| SAFE-LSTMPolicy=π0, Category=Class., Venue=NeurIPS'252026.05 | 77.4 | 85.1 | 66.3 | |
| Cosine k-NNPolicy=OpenVLA, Category=OOD Det., Venue=ICML'222026.05 | 75.8 | 75.7 | 63.2 | |
| RNDPolicy=π0, Category=OOD Det., Venue=ICML'242026.05 | 70.6 | 78.4 | 62.9 | |
| EigenScorePolicy=π0, Category=Multi., Venue=ICLR'242026.05 | 69.1 | 76.1 | 53.4 | |
| ACEPolicy=π0, Category=Multi., Venue=NeurIPS'252026.05 | 67.2 | 76.4 | 57.2 | |
| MahalanobisPolicy=OpenVLA, Category=OOD Det., Venue=ICML'182026.05 | 67 | 66.7 | 60.4 | |
| SAFE-LSTMPolicy=OpenVLA, Category=Class., Venue=NeurIPS'252026.05 | 67 | 66.5 | 59.3 | |
| STACPolicy=OpenVLA, Category=Multi., Venue=CoRL'24, note=Adapted to single-step output2026.05 | 66.5 | 65.1 | 54.5 | |
| EigenScorePolicy=OpenVLA, Category=Multi., Venue=ICLR'242026.05 | 66.3 | 63.6 | 53.6 | |
| LogpZOPolicy=OpenVLA, Category=OOD Det., Venue=RSS'252026.05 | 65.8 | 66.5 | 59.4 | |
| STACPolicy=π0, Category=Multi., Venue=CoRL'242026.05 | 61.9 | 75.1 | 50.1 | |
| ACEPolicy=OpenVLA, Category=Multi., Venue=NeurIPS'252026.05 | 60.7 | 63.1 | 51.6 | |
| PCA KMeansPolicy=π0, Category=OOD Det., Venue=CoRL'242026.05 | 59.1 | 59.5 | 47.6 | |
| Cosine k-NNPolicy=π0, Category=OOD Det., Venue=ICML'222026.05 | 58 | 84.1 | 56.5 | |
| Cluster EntropyPolicy=OpenVLA, Category=Multi., Venue=ICLR'232026.05 | 55.2 | 52.4 | 51.6 | |
| Cluster EntropyPolicy=π0, Category=Multi., Venue=ICLR'232026.05 | 54.8 | 74.9 | 45.4 | |
| PCA KMeansPolicy=OpenVLA, Category=OOD Det., Venue=CoRL'242026.05 | 53.4 | 54.4 | 49.4 | |
| RNDPolicy=OpenVLA, Category=OOD Det., Venue=ICML'242026.05 | 52.6 | 52.3 | 49.3 | |
| Max EntropyPolicy=OpenVLA, Category=Token Unc., Venue=ICLR'212026.05 | 51.4 | 52.7 | 45.5 | |
| Avg. EntropyPolicy=OpenVLA, Category=Token Unc., Venue=ICLR'212026.05 | 51.3 | 52.4 | 47.8 | |
| Avg. NLLPolicy=OpenVLA, Category=Token Unc., Venue=ICLR'232026.05 | 50.6 | 52.2 | 47.5 | |
| Max NLLPolicy=OpenVLA, Category=Token Unc., Venue=ICLR'232026.05 | 49.4 | 49.3 | 45.6 |