ResearchBenchmarksNoisy Action Robust MDP on Walker2D MuJoCo (low randomness)Follow429.6Worst-case ReturnDT378.64391.87405.1418.33Jul 25, 2024Evaluation ResultsMethodMethodLinksWorst-case ReturnDTnumber of seeds=5, adv...number of seeds=5, adversarial policies=8, target returns=best2024.07429.6ARDTnumber of seeds=5, adv...number of seeds=5, adversarial policies=8, target returns=best2024.07405.6ESPERnumber of seeds=5, adv...number of seeds=5, adversarial policies=8, target returns=best2024.07380.6