JAEGER: Dual-Level Humanoid Whole-Body Controller

TL;DR

JAEGER introduces a dual-level control framework for humanoid robots, separating upper and lower body control to enhance robustness, versatility, and fault tolerance, trained via supervised and reinforcement learning on human motion data.

Contribution

The paper proposes a novel dual-level control architecture for humanoids, improving stability and task focus by separating control of upper and lower bodies, trained with a curriculum learning approach.

Findings

Outperforms state-of-the-art methods in simulation and real environments.

Enables versatile movements through coarse and fine control levels.

Improves fault tolerance and robustness in humanoid control.

Abstract

This paper presents JAEGER, a dual-level whole-body controller for humanoid robots that addresses the challenges of training a more robust and versatile policy. Unlike traditional single-controller approaches, JAEGER separates the control of the upper and lower bodies into two independent controllers, so that they can better focus on their distinct tasks. This separation alleviates the dimensionality curse and improves fault tolerance. JAEGER supports both root velocity tracking (coarse-grained control) and local joint angle tracking (fine-grained control), enabling versatile and stable movements. To train the controller, we utilize a human motion dataset (AMASS), retargeting human poses to humanoid poses through an efficient retargeting network, and employ a curriculum learning approach. This method performs supervised learning for initialization, followed by reinforcement learning for further exploration. We conduct our experiments on two humanoid platforms and demonstrate the superiority of our approach against state-of-the-art methods in both simulation and real environments.