Share your thoughts, 1 month free Claude Pro on usSee more

Rollout-level failure detection on ReactorX arm SmolVLA policy rollouts

0.82ROC-AUC

RND

Updated 1mo ago

Evaluation Results

Method	Links
RND 2026.06		0.82
Foresight-Transformer 2026.06		0.79
FAIL-Detect 2026.06		0.71
SAFE-MLP 2026.06		0.64
Foresight-LSTM 2026.06		0.64
Foresight-MLP 2026.06		0.53
SAFE-LSTM 2026.06		0.43