List your company

Failure Detection on WidowX (unseen)

0.153Brier ScoreSAFE-RNN-TDQC (Ours)
0.135960.250980.3660.48102Apr 22, 2026

Evaluation results

MethodLinks
SAFE-RNN-TDQC (Ours)VLA Model=OpenVLA2026.040.153
SAFE-MLP BCEVLA Model=OpenVLA2026.040.164
SAFE-MLP-TDQC (Ours)VLA Model=OpenVLA2026.040.169
RNN-TDQC (Ours)VLA Model=OpenVLA2026.040.192
SAFE-RNNVLA Model=OpenVLA2026.040.213
Running Avg prob.VLA Model=OpenVLA2026.040.257
Avg prob.VLA Model=OpenVLA2026.040.282
RNN-BCEVLA Model=OpenVLA2026.040.344
Avg entropyVLA Model=OpenVLA2026.040.426
Running Avg entropyVLA Model=OpenVLA2026.040.432
Max prob.VLA Model=OpenVLA2026.040.579