| title | A Safe Action Is Not Enough: Feasible-Future Decoding for Vision-Language-Action Policies |
|---|---|
| authors | Tu Nguyen, Matthieu Zimmer, Vu Anh Vu, Ziyi Wang, Jannik Hammel Nielsen, Xuebing Zhou, Haitham Bou Ammar |
| date | 2026-10-06 |
| venue | arXiv (preprint, not peer reviewed) |
| id | arXiv:2610.05166 |
| affiliation | HUAWEI Noah's Ark Lab |
| upvotes | 3 (Hugging Face Daily Papers, 2026-11-01) |
| links | abs pdf hf |
| primer | Vision-language-action models |
| abstract | A safe action is not necessarily a viable one. A frozen vision-language-action (VLA) policy can favor a locally admissible move that leaves no policy-supported route to safe task completion. We call this the feasibility-likelihood gap: likelihood ranks the next move, while feasibility depends on the futures it leaves open. To bring those futures into the decision, we derive the exact next-block marginal of the history-conditioned policy-environment trajectory law restricted to safe task completion. The derivation reveals a candidate-dependent feasible-future mass: its support records whether safe completion remains possible under the frozen continuation process, while its magnitude measures how much weighted safe-completion mass remains. Since exact evaluation is impractical online, we develop a selective finite-candidate approximation and establish conditions for recovering the best retained viable candidate. Our alarm-triggered, training-free reranker VICS-G lowers mean cumulative safety cost by 1.9%-57.5% across six Safety-CHORES settings while remaining within 2.5 percentage points of policy sampling in success and 0.82 steps in mean episode length. Our approach offers a promising and practical path toward safer task completion, grounded in an exact policy-relative target yet requiring neither policy retraining nor online rollouts. |
cite
@misc{nguyen2026safe,
title={A Safe Action Is Not Enough: Feasible-Future Decoding for Vision-Language-Action Policies},
author={Tu Nguyen and Matthieu Zimmer and Vu Anh Vu and Ziyi Wang and Jannik Hammel Nielsen and Xuebing Zhou and Haitham Bou Ammar},
year={2026},
eprint={2610.05166},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2610.05166}
}