Distilling Motion Planner Augmented Policies into Visual Control Policies for Robot Manipulation
This addresses the problem of deploying robot manipulation policies in real-world settings without state information, offering a solution for robotics researchers and practitioners, though it is incremental as it builds on prior integration of motion planning and reinforcement learning.
The paper tackles the challenge of learning complex robot manipulation tasks in obstructed environments by distilling a state-based motion planner augmented policy into a visual control policy, achieving high sample efficiency and outperforming state-of-the-art algorithms in three manipulation tasks.
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