ResearchBenchmarksContinuous Control on pendulumFollow0.946Average RewardDDPG0.065120.293810.52250.75119Sep 9, 2015Evaluation ResultsMethodMethodLinksAverage RewardBest RewardDDPGObservation Space=low-...Observation Space=low-dimensional2015.090.9461.021DDPGObservation Space=pixelsObservation Space=pixels2015.090.6631.055DPGSetup=original with re...Setup=original with replay buffer and batch normalization2015.090.0990.951