Robust Recovery Motion Control for Quadrupedal Robots via Learned Terrain Imagination

TL;DR

This paper introduces a robust recovery control policy for quadrupedal robots that leverages learned terrain imagination to enhance recovery in cluttered environments, validated through simulations and real-world tests.

Contribution

It presents a novel all-terrain recovery policy specifically designed for cluttered environments, improving upon existing deep reinforcement learning approaches.

Findings

Superior recovery performance in simulations

Effective real-world recovery in diverse terrains

Enhanced safety and speed in recovery tasks

Abstract

Quadrupedal robots have emerged as a cutting-edge platform for assisting humans, finding applications in tasks related to inspection and exploration in remote areas. Nevertheless, their floating base structure renders them susceptible to fall in cluttered environments, where manual recovery by a human operator may not always be feasible. Several recent studies have presented recovery controllers employing deep reinforcement learning algorithms. However, these controllers are not specifically designed to operate effectively in cluttered environments, such as stairs and slopes, which restricts their applicability. In this study, we propose a robust all-terrain recovery policy to facilitate rapid and secure recovery in cluttered environments. We substantiate the superiority of our proposed approach through simulations and real-world tests encompassing various terrain types.