Learning Bipedal Walking on a Quadruped Robot via Adversarial Motion Priors

TL;DR

This paper adapts a learning framework with adversarial motion priors to enable a quadruped robot to perform stable bipedal walking in complex terrains, including stairs and uneven surfaces, demonstrating successful transfer and navigation capabilities.

Contribution

It introduces a novel application of adversarial motion priors for bipedal locomotion on quadruped robots, extending previous quadrupedal learning methods to bipedal modes.

Findings

Successful transfer of the learning framework to bipedal mode

Enables navigation on flat, stairs, and uneven terrains

Demonstrates stable walking in simulation environments

Abstract

Previous studies have successfully demonstrated agile and robust locomotion in challenging terrains for quadrupedal robots. However, the bipedal locomotion mode for quadruped robots remains unverified. This paper explores the adaptation of a learning framework originally designed for quadrupedal robots to operate blind locomotion in biped mode. We leverage a framework that incorporates Adversarial Motion Priors with a teacher-student policy to enable imitation of a reference trajectory and navigation on tough terrain. Our work involves transferring and evaluating a similar learning framework on a quadruped robot in biped mode, aiming to achieve stable walking on both flat and complicated terrains. Our simulation results demonstrate that the trained policy enables the quadruped robot to navigate both flat and challenging terrains, including stairs and uneven surfaces.