ResearchBenchmarksMulti-Objective Reinforcement Learning on mo-halfcheetah v5Multi-Objective Reinforcement Learning on mo-halfcheetah v5Follow22,500HV (x10^4)PRISM2446,02211,80017,578Feb 20, 2026Feb 21, 2026Feb 23, 2026Feb 25, 2026Feb 27, 2026Mar 1, 2026Mar 3, 2026Evaluation resultsMethodMethodLinksHV (x10^4)EUMVOSparsityPRISMReward Sparsity=Extrem...Reward Sparsity=Extreme (first objective)2026.0222,50089.9440.72—OracleReward Sparsity=None (...Reward Sparsity=None (Dense)2026.0217,00081.2936.84—BaselineReward Sparsity=Extrem...Reward Sparsity=Extreme (first objective)2026.029,700-1.46-1.01—MORL/D2026.031,880——136.4MOPDERL2026.031,550——101.7MAPEX2026.031,100——337.5