Benchmarking Potential Based Rewards for Learning Humanoid Locomotion | IEEE Conference Publication | IEEE Xplore