Sim-to-Real Transfer of Robot Learning with Variable Length Inputs
This addresses the challenge of slow training times and lack of prior knowledge incorporation in RL for robotics, enabling faster sim-to-real transfer for tasks like object sorting, though it is incremental as it builds on existing modular RL and deep sets techniques.
The paper tackles the problem of training deep reinforcement learning policies for robotics with variable-length inputs, which is a bottleneck for sim-to-real transfer, by proposing a framework that combines deep sets encoding with modular RL. The result is a method that learns effective policies in minutes of simplified simulation and directly transfers to a real robot, generalizing to unseen task variations.
Current end-to-end deep Reinforcement Learning (RL) approaches require jointly learning perception, decision-making and low-level control from very sparse reward signals and high-dimensional inputs, with little capability of incorporating prior knowledge. This results in prohibitively long training times for use on real-world robotic tasks. Existing algorithms capable of extracting task-level representations from high-dimensional inputs, e.g. object detection, often produce outputs of varying lengths, restricting their use in RL methods due to the need for neural networks to have fixed length inputs. In this work, we propose a framework that combines deep sets encoding, which allows for variable-length abstract representations, with modular RL that utilizes these representations, decoupling high-level decision making from low-level control. We successfully demonstrate our approach on the robot manipulation task of object sorting, showing that this method can learn effective policies within mere minutes of highly simplified simulation. The learned policies can be directly deployed on a robot without further training, and generalize to variations of the task unseen during training.