Reinforced Grounded Action Transformation for Sim-to-Real Transfer

TL;DR

This paper introduces Reinforced Grounded Action Transformation (RGAT), a novel RL-based method that improves sim-to-real transfer by end-to-end training of the grounding process, outperforming previous methods like GAT in complex neural network policies.

Contribution

The paper presents RGAT, a new RL-based approach that enhances the grounding step in sim-to-real transfer, enabling more robust transfer for complex neural network policies.

Findings

RGAT outperforms GAT in MuJoCo domains.

End-to-end training improves the grounded simulator quality.

Successful transfer achieved for neural network policies.

Abstract

Robots can learn to do complex tasks in simulation, but often, learned behaviors fail to transfer well to the real world due to simulator imperfections (the reality gap). Some existing solutions to this sim-to-real problem, such as Grounded Action Transformation (GAT), use a small amount of real-world experience to minimize the reality gap by grounding the simulator. While very effective in certain scenarios, GAT is not robust on problems that use complex function approximation techniques to model a policy. In this paper, we introduce Reinforced Grounded Action Transformation(RGAT), a new sim-to-real technique that uses Reinforcement Learning (RL) not only to update the target policy in simulation, but also to perform the grounding step itself. This novel formulation allows for end-to-end training during the grounding step, which, compared to GAT, produces a better grounded simulator. Moreover, we show experimentally in several MuJoCo domains that our approach leads to successful transfer for policies modeled using neural networks.