# Towards Safe Robot Foundation Models Using Inductive Biases

arXiv: https://arxiv.org/abs/2505.10219 · HTML: https://mlsanity.com/paper/2505.10219/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 15 May 2025.

## Built on by (most cited first, 5 of 5)

- [Vision-Language-Action Safety: Threats, Challenges, Evaluations, and Mechanisms](https://mlsanity.com/paper/2604.23775/) (2026)
- [Operator Models for Continuous-Time Offline Reinforcement Learning](https://mlsanity.com/paper/2511.10383/) (2025)
- [How VLAs Fail Differently: Black-Box Action Monitoring Reveals Architecture-Specific Failure Signatures](https://mlsanity.com/paper/2605.28726/) (2026)
- [Security of Foundation-Model-Powered Embodied Agents: Attack Surfaces, Attacks, Defenses, and Evaluation](https://mlsanity.com/paper/2608.16843/) (2026)
- [Security and Privacy in Large-Model-Driven Embodied Agents: Attacks, Defenses, and Future Directions](https://mlsanity.com/paper/2609.27847/) (2026)

## Abstract (arXiv)

Safety is a critical requirement for the real-world deployment of robotic systems. Unfortunately, while current robot foundation models show promising generalization capabilities across a wide variety of tasks, they fail to address safety, an important aspect for ensuring long-term operation. Current robot foundation models assume that safe behavior should emerge by learning from a sufficiently large dataset of demonstrations. However, this approach has two clear major drawbacks. Firstly, there are no formal safety guarantees for a behavior cloning policy trained using supervised learning. Secondly, without explicit knowledge of any safety constraints, the policy may require an unreasonable number of additional demonstrations to even approximate the desired constrained behavior. To solve these key issues, we show how we can instead combine robot foundation models with geometric inductive biases using ATACOM, a safety layer placed after the foundation policy that ensures safe state transitions by enforcing action constraints. With this approach, we can ensure formal safety guarantees for generalist policies without providing extensive demonstrations of safe behavior, and without requiring any specific fine-tuning for safety. Our experiments show that our approach can be beneficial both for classical manipulation tasks, where we avoid unwanted collisions with irrelevant objects, and for dynamic tasks, such as the robot air hockey environment, where we can generate fast trajectories respecting complex tasks and joint space constraints.