Deep RL for autonomous robots: limitations and safety challenges.

摘要

With the rise of deep reinforcement learning, there has also been a string of successes on continuous control problems using physics simulators. This has lead to some optimism regarding use in auto ...