Learning a Centroidal Motion Planner for Legged Locomotion

TL;DR

This paper introduces a neural network-based method to generate real-time centroidal motions for legged robots, enabling fast and dynamic locomotion behaviors like walking and jumping, trained using existing optimizers.

Contribution

It presents a novel approach to learn a centroidal motion predictor that allows real-time whole-body motion generation for legged robots, bridging the gap between offline optimization and online control.

Findings

Enables real-time generation of complex locomotion behaviors.

Achieves high-frequency control suitable for practical deployment.

Demonstrates successful walking and jumping motions on a real robot.

Abstract

Whole-body optimizers have been successful at automatically computing complex dynamic locomotion behaviors. However they are often limited to offline planning as they are computationally too expensive to replan with a high frequency. Simpler models are then typically used for online replanning. In this paper we present a method to generate whole body movements in real-time for locomotion tasks. Our approach consists in learning a centroidal neural network that predicts the desired centroidal motion given the current state of the robot and a desired contact plan. The network is trained using an existing whole body motion optimizer. Our approach enables to learn with few training samples dynamic motions that can be used in a complete whole-body control framework at high frequency, which is usually not attainable with typical full-body optimizers. We demonstrate our method to generate a rich set of walking and jumping motions on a real quadruped robot.