ResearchBenchmarksSafe Reinforcement Learning on DSRL EasyDenseFollow316.47RewardCDCP55.2012123.0306190.86258.6894Jul 4, 2026Evaluation ResultsMethodMethodLinksRewardCostCDCP2026.07316.4724.39CMTDiff2026.07308.4537.2CMTDD2026.0765.2522.33