JAEGER: Dual-Level Humanoid Whole-Body Controller

2025-05-10

AI Summary

This work addresses the poor robustness and limited task adaptability in whole-body control of humanoid robots, arising from strong upper-lower body coupling. We propose a decoupled two-level whole-body controller. Methodologically, we introduce a novel dual-controller architecture that explicitly separates upper- and lower-body control, integrated with a hybrid training paradigm combining kinematic retargeting and curriculum learning: supervised pretraining is performed using a pose retargeting network on the AMASS dataset, followed by reinforcement learning fine-tuning with dual objectives—root velocity and joint angle tracking. This design effectively mitigates the curse of dimensionality in high-dimensional control spaces, significantly enhancing fault tolerance and motion flexibility. Extensive evaluations in simulation and on real humanoid platforms demonstrate superior performance over existing state-of-the-art methods in motion stability, behavioral diversity, and cross-task generalization.

Technology Category

• Intelligent Robots: Behavior Learning & Control
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Natural Language Processing: Safety and Robustness

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

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.

Problem

• Develops dual-level controller for humanoid robots
• Separates upper and lower body control
• Enables versatile and stable movements

Innovation

• Dual-level controller separates upper and lower bodies
• Uses human motion dataset for retargeting poses
• Combines supervised and reinforcement learning methods