Distilling Motion Planner Augmented Policies into Visual Control Policies for Robot Manipulation

TL;DR

This paper introduces a method to convert motion planner augmented policies into visual control policies for robot manipulation, enabling efficient learning and zero-shot transfer in obstructed environments.

Contribution

The authors propose a novel distillation approach combining visual behavioral cloning and vision-based reinforcement learning to create effective visual policies from state-based motion planner policies.

Findings

The method is highly sample-efficient and outperforms existing algorithms.

It enables zero-shot transfer to unseen environments with distractors.

The approach successfully handles complex manipulation tasks in obstructed settings.

Abstract

Learning complex manipulation tasks in realistic, obstructed environments is a challenging problem due to hard exploration in the presence of obstacles and high-dimensional visual observations. Prior work tackles the exploration problem by integrating motion planning and reinforcement learning. However, the motion planner augmented policy requires access to state information, which is often not available in the real-world settings. To this end, we propose to distill a state-based motion planner augmented policy to a visual control policy via (1) visual behavioral cloning to remove the motion planner dependency along with its jittery motion, and (2) vision-based reinforcement learning with the guidance of the smoothed trajectories from the behavioral cloning agent. We evaluate our method on three manipulation tasks in obstructed environments and compare it against various reinforcement learning and imitation learning baselines. The results demonstrate that our framework is highly sample-efficient and outperforms the state-of-the-art algorithms. Moreover, coupled with domain randomization, our policy is capable of zero-shot transfer to unseen environment settings with distractors. Code and videos are available at https://clvrai.com/mopa-pd