Real World Reinforcement Learning of Active Perception Behaviors

A robot's instantaneous sensory observations do not always reveal task-relevant state information. Under such partial observability, optimal behavior typically involves explicitly acting to gain the missing information. Today's standard robot learning techniques struggle to produce such active perception behaviors. We propose a simple real-world robot learning recipe to efficiently train active perception policies. Our approach, advantage weighted regression (AAWR), exploits access to "privileged" extra sensors at training time. The privileged sensors enable training high-quality privileged value functions that aid in estimating the advantage of the target policy. Bootstrapping from a small number of potentially suboptimal demonstrations and an easy-to-obtain coarse policy initialization, AAWR quickly acquires active perception behaviors and boosts task performance. In evaluations on 8 manipulation tasks on 3 robots spanning varying degrees of partial observability, AAWR synthesizes reliable active perception behaviors that outperform all prior approaches. When initialized with a "generalist" robot policy that struggles with active perception tasks, AAWR efficiently generates information-gathering behaviors that allow it to operate under severe partial observability for manipulation tasks. Website:penn-pal-lab.github.io/aawr

Our method, Asymmetric Advantage Weighted Regression (AAWR), uses privileged information to learn high-quality value functions. The value functions are used to compute advantage estimates, which are used as weights in the weighted BC loss, leading to better policy extraction in POMDPs compared to BC and unprivileged AWR. AAWR inherits the versatility of AWR, which can be applied to both offline data (demonstrations, play data) and online data using the same update.