ResearchBenchmarksSafe Reinforcement Learning on single-gate Differential Drive dynamicsFollow0.1Safety Violations / Episode (Mean)TD3-0.1641.6183.45.182May 22, 2024Evaluation ResultsMethodMethodLinksSafety Violations / Episode (Mean)Shield Invocations / Episode (Mean)Shield Invocations / Episode (SD)Safety Violations / Episode (SD)TD32024.050.1——0.2CPO2024.055.2——3.5PPO-lag2024.056.7——7.2DMPS2024.05—3.11.6—MPS2024.05—7.46.9—