Publication
Towards Safe Robot Foundation Models
Maximilian Tölle; Theo Gruner; Daniel Palenicek; Jonas Günster; Puze Liu; Joe Watson; Davide Tateo; Jan Peters
In: Computing Research Repository eprint Journal (CoRR), Vol. abs/2503.07404, Pages 1-3, arXiv, 2025.
Abstract
Robot foundation models hold the potential for
deployment across diverse environments, from industrial ap-
plications to household tasks. While current research focuses
primarily on the policies’ generalization capabilities across
a variety of tasks, it fails to address safety, a critical re-
quirement for deployment on real-world systems. In this
paper, we introduce a safety layer designed to constrain
the action space of any generalist policy appropriately. Our
approach uses ATACOM, a safe reinforcement learning algo-
rithm that creates a safe action space and, therefore, ensures
safe state transitions. By extending ATACOM to generalist
policies, our method facilitates their deployment in safety-
critical scenarios without requiring any specific safety fine-
tuning. We demonstrate the effectiveness of this safety layer
in an air hockey environment, where it prevents a puck-
hitting agent from colliding with its surroundings, a failure
observed in generalist policies. https://sites.google.
com/robot-learning.de/towards-safe-rfm