Risk-Averse Model Uncertainty for Distributionally Robust Safe Reinforcement Learning
This work addresses safety in reinforcement learning for real-world applications, offering an efficient, model-free approach without minimax optimization, though it appears incremental as it builds on existing distributionally robust methods.
The authors tackled safe decision-making in uncertain environments by introducing a deep reinforcement learning framework that uses coherent distortion risk measures to handle model uncertainty, achieving robust performance and safety across perturbed test environments in continuous control tasks.
Many real-world domains require safe decision making in uncertain environments. In this work, we introduce a deep reinforcement learning framework for approaching this important problem. We consider a distribution over transition models, and apply a risk-averse perspective towards model uncertainty through the use of coherent distortion risk measures. We provide robustness guarantees for this framework by showing it is equivalent to a specific class of distributionally robust safe reinforcement learning problems. Unlike existing approaches to robustness in deep reinforcement learning, however, our formulation does not involve minimax optimization. This leads to an efficient, model-free implementation of our approach that only requires standard data collection from a single training environment. In experiments on continuous control tasks with safety constraints, we demonstrate that our framework produces robust performance and safety at deployment time across a range of perturbed test environments.