Domain Randomization via Entropy Maximization

arXiv:2311.01885v2 Announce Type: replace
Abstract: Varying dynamics parameters in simulation is a popular Domain Randomization (DR) approach for overcoming the reality gap in Reinforcement Learning (RL). Nevertheless, DR heavily hinges on the choice of the sampling distribution of the dynamics parameters, since high variability is crucial to regularize the agent's behavior but notoriously leads to overly conservative policies when randomizing excessively. In this paper, we propose a novel approach to address sim-to-real transfer, which automatically shapes dynamics distributions during training …

abstract agent arxiv behavior cs.lg cs.ro distribution domain dynamics entropy gap leads parameters popular randomization reality reinforcement reinforcement learning sampling simulation type via