LiRA: Light-Robust Adversary for Model-based Reinforcement Learning in Real World

TL;DR

LiRA introduces a novel adversarial learning framework for model-based reinforcement learning that balances robustness and performance, enabling effective real-world robotic control with limited data.

Contribution

The paper proposes LiRA, a new adversarial learning approach that automatically adjusts robustness levels using variational inference and a light robustness constraint.

Findings

LiRA effectively balances robustness and control performance.

LiRA enables quadrupedal robots to learn force-reactive gait control with less than two hours of real-world data.

Numerical simulations confirm LiRA's robustness and adaptability.

Abstract

Model-based reinforcement learning has attracted much attention due to its high sample efficiency and is expected to be applied to real-world robotic applications. In the real world, as unobservable disturbances can lead to unexpected situations, robot policies should be taken to improve not only control performance but also robustness. Adversarial learning is an effective way to improve robustness, but excessive adversary would increase the risk of malfunction, and make the control performance too conservative. Therefore, this study addresses a new adversarial learning framework to make reinforcement learning robust moderately and not conservative too much. To this end, the adversarial learning is first rederived with variational inference. In addition, \textit{light robustness}, which allows for maximizing robustness within an acceptable performance degradation, is utilized as a constraint. As a result, the proposed framework, so-called LiRA, can automatically adjust adversary level, balancing robustness and conservativeness. The expected behaviors of LiRA are confirmed in numerical simulations. In addition, LiRA succeeds in learning a force-reactive gait control of a quadrupedal robot only with real-world data collected less than two hours.