ResearchBenchmarksSafe Reinforcement Learning on Safety Gymnasium Walker2d (threshold=5)Safe Reinforcement Learning on Safety Gymnasium Walker2d (threshold=5)Follow3,057ReturnPPO-Lag2,389.322,562.662,7362,909.34Mar 24, 2026Evaluation resultsMethodMethodLinksReturnCostPPO-LagType=OracleType=Oracle2026.033,0574.88PPO-BT2026.032,8074.91PbCRL2026.032,6594.62RLSF2026.032,4153.66