Go to NeurIPS 2025 Conference homepageSafeVLA: Towards Safety Alignment of Vision-Language-Action Model via Constrained Learning
Keywords: Vision-Language-Action Models, Safety Alignment, Large-Scale Constrained Learning
TL;DR: We make Vision-Language-Action models (VLAs) significantly safer (83.58% improvement, no performance loss) by explicitly integrating constraints via a new integrated safety approach (ISA) based on CMDPs/SafeRL, validated on our new benchmark.
Abstract: Vision-language-action models (VLAs) show potential as generalist robot policies. However, these models pose extreme safety challenges during real-world deployment, including the risk of harm to the environment, the robot itself, and humans. *How can safety constraints be explicitly integrated into VLAs?* We address this by exploring an integrated safety approach (ISA), systematically **modeling** safety requirements, then actively **eliciting** diverse unsafe behaviors, effectively **constraining** VLA policies via safe reinforcement learning, and rigorously **assuring** their safety through targeted evaluations. Leveraging the constrained Markov decision process (CMDP) paradigm, ISA optimizes VLAs from a min-max perspective against elicited safety risks. Thus, policies aligned through this comprehensive approach achieve the following key features: (I) effective **safety-performance trade-offs**, reducing the cumulative cost of safety violations by 83.58\% compared to the state-of-the-art method, while also maintaining task success rate (+3.85\%). (II) strong **safety assurance**, with the ability to mitigate long-tail risks and handle extreme failure scenarios. (III) robust **generalization** of learned safety behaviors to various out-of-distribution perturbations. The effectiveness is evaluated on long-horizon mobile manipulation tasks.
Primary Area: Social and economic aspects of machine learning (e.g., fairness, interpretability, human-AI interaction, privacy, safety, strategic behavior)
Submission Number: 19550
Loading