Agile and Versatile Robot Locomotion via Kernel-based Residual Learning

TL;DR

This paper introduces a kernel-based residual learning framework for quadrupedal robots, enabling robust, versatile, and efficient omnidirectional locomotion across various terrains and gaits, surpassing traditional MPC controllers.

Contribution

It presents a novel kernel residual learning approach that enhances quadrupedal locomotion robustness and versatility with high sample efficiency and generalization capabilities.

Findings

Achieves omnidirectional locomotion at continuous velocities

Successfully generalizes to unseen terrains and gaits

Demonstrates high sample efficiency and controllability

Abstract

This work developed a kernel-based residual learning framework for quadrupedal robotic locomotion. Initially, a kernel neural network is trained with data collected from an MPC controller. Alongside a frozen kernel network, a residual controller network is trained via reinforcement learning to acquire generalized locomotion skills and resilience against external perturbations. With this proposed framework, a robust quadrupedal locomotion controller is learned with high sample efficiency and controllability, providing omnidirectional locomotion at continuous velocities. Its versatility and robustness are validated on unseen terrains that the expert MPC controller fails to traverse. Furthermore, the learned kernel can produce a range of functional locomotion behaviors and can generalize to unseen gaits.