How to pick the domain randomization parameters for sim-to-real transfer of reinforcement learning policies?

TL;DR

This paper investigates how to optimally select domain randomization parameters in simulation to improve the transfer of reinforcement learning policies to real-world robotics, highlighting the importance of distribution choice.

Contribution

It demonstrates that the distribution parameters significantly impact real-world performance and proposes methods to optimize these parameters for better sim-to-real transfer.

Findings

Distribution choice greatly affects real-world policy performance.

Optimizing distribution parameters improves transfer success.

Hand-tuned distributions are less effective than optimized ones.

Abstract

Recently, reinforcement learning (RL) algorithms have demonstrated remarkable success in learning complicated behaviors from minimally processed input. However, most of this success is limited to simulation. While there are promising successes in applying RL algorithms directly on real systems, their performance on more complex systems remains bottle-necked by the relative data inefficiency of RL algorithms. Domain randomization is a promising direction of research that has demonstrated impressive results using RL algorithms to control real robots. At a high level, domain randomization works by training a policy on a distribution of environmental conditions in simulation. If the environments are diverse enough, then the policy trained on this distribution will plausibly generalize to the real world. A human-specified design choice in domain randomization is the form and parameters of the distribution of simulated environments. It is unclear how to the best pick the form and parameters of this distribution and prior work uses hand-tuned distributions. This extended abstract demonstrates that the choice of the distribution plays a major role in the performance of the trained policies in the real world and that the parameter of this distribution can be optimized to maximize the performance of the trained policies in the real world