Privileged observations enable rapid and reliable policy discovery directly in the physical world

Antonio Terpin, Raffaello D'Andrea

We study how privileged information about a physical system affects the discovery of high-performing policies when training a reinforcement learning agent directly in the physical…