Asymmetric Actor Critic for Image-Based Robot Learning

TL;DR

This paper introduces an asymmetric actor-critic reinforcement learning approach for image-based robot control, leveraging full state information in simulation to improve policies that operate on partial observations, enabling effective sim-to-real transfer without real data.

Contribution

It proposes an asymmetric actor-critic method that trains the critic on full states while the actor uses only partial observations, enhancing sim-to-real transfer in robotic tasks.

Findings

Asymmetric training improves policy performance in simulation.

Method enables successful transfer to real robots without real data.

Combining with domain randomization enhances robustness.

Abstract

Deep reinforcement learning (RL) has proven a powerful technique in many sequential decision making domains. However, Robotics poses many challenges for RL, most notably training on a physical system can be expensive and dangerous, which has sparked significant interest in learning control policies using a physics simulator. While several recent works have shown promising results in transferring policies trained in simulation to the real world, they often do not fully utilize the advantage of working with a simulator. In this work, we exploit the full state observability in the simulator to train better policies which take as input only partial observations (RGBD images). We do this by employing an actor-critic training algorithm in which the critic is trained on full states while the actor (or policy) gets rendered images as input. We show experimentally on a range of simulated tasks that using these asymmetric inputs significantly improves performance. Finally, we combine this method with domain randomization and show real robot experiments for several tasks like picking, pushing, and moving a block. We achieve this simulation to real world transfer without training on any real world data.