RobustRLlib: A Unified Library and Benchmark for Robust Reinforcement Learning Algorithms.

arXiv:2610.32971v1 Announce Type: cross Abstract: Robust reinforcement learning (RL) is critical for reliable real-world deployment, where policies may fail under diverse shifts and disruptions. Existing robust RL algorithms offer a rich set of mechanisms and