# VLA-Stem: Adding a Lightweight Safety Mechanism to Vision-Language-Action Models with Reinforcement Learning

- Source: arXiv cs.RO
- Published: 2026-10-28T04:00:00.000Z
- RoboHot: https://robohot.news/en/items/bakfxyrl3fyje5k8fu9ps33kj
- Original: https://arxiv.org/abs/2610.37378

## Summary

Researchers propose VLA-Stem, a lightweight reinforcement-learning-based safety mechanism that does not fine-tune the underlying VLA policy; instead, it converts its action chunks into motion references and, at inference time, adaptively corrects execution using proprioception and local obstacle information. Across six constrained simulation settings, VLA-Stem achieved an average collision-free task success rate of 52.8%, versus 23.7% for the strongest safety baseline, while preserving the original policy's performance in obstacle-free environments; real-robot experiments also showed reduced collisions under static and dynamic obstacles.