Improve Robustness of Reinforcement Learning against Observation Perturbations via $l_\infty$ Lipschitz Policy Networks

no code implementations • 14 Dec 2023 • Buqing Nie, Jingtian Ji, Yangqing Fu, Yue Gao

In this work, we propose a novel robust reinforcement learning method called SortRL, which improves the robustness of DRL policies against observation perturbations from the perspective of the network architecture.

reinforcement-learning