Assessing Transferability from Simulation to Reality for Reinforcement Learning
This work addresses a critical problem in robotics for enabling safer and more efficient policy learning by reducing reliance on expensive real-world experiments.
The paper tackled the challenge of transferring reinforcement learning policies from simulation to reality by addressing the Simulation Optimization Bias (SOB) that can cause damage to robots. They proposed the SPOTA algorithm with domain randomization and a stopping criterion based on SOB estimation, achieving direct policy transfer to real systems without additional training.
Learning robot control policies from physics simulations is of great interest to the robotics community as it may render the learning process faster, cheaper, and safer by alleviating the need for expensive real-world experiments. However, the direct transfer of learned behavior from simulation to reality is a major challenge. Optimizing a policy on a slightly faulty simulator can easily lead to the maximization of the `Simulation Optimization Bias` (SOB). In this case, the optimizer exploits modeling errors of the simulator such that the resulting behavior can potentially damage the robot. We tackle this challenge by applying domain randomization, i.e., randomizing the parameters of the physics simulations during learning. We propose an algorithm called Simulation-based Policy Optimization with Transferability Assessment (SPOTA) which uses an estimator of the SOB to formulate a stopping criterion for training. The introduced estimator quantifies the over-fitting to the set of domains experienced while training. Our experimental results on two different second order nonlinear systems show that the new simulation-based policy search algorithm is able to learn a control policy exclusively from a randomized simulator, which can be applied directly to real systems without any additional training.