Asymmetric Actor Critic for Image-Based Robot Learning
This work addresses the problem of expensive and dangerous physical training for robotics by enabling more efficient simulation-to-real transfer, though it is incremental as it builds on existing actor-critic and domain randomization methods.
The authors tackled the challenge of training robot control policies in simulation for real-world transfer by exploiting full state observability in simulators to train policies using only partial RGBD image inputs, achieving significant performance improvements on simulated tasks and successfully transferring to real robots without real-world data.
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.