EMP: Executable Motion Prior for Humanoid Robot Standing Upper-body Motion Imitation

2025-07-21

AI Summary

This work addresses the challenge of high-fidelity upper-body motion imitation by humanoid robots while maintaining stable bipedal standing. We propose a reinforcement learning–based whole-body coordination control framework. Its core innovation is the Executable Motion Prior (EMP) module, which dynamically refines target poses based on real-time robot state, preserving motion amplitude while substantially improving motion safety and standing stability. To enable robust policy learning, we employ a motion retargeting network to synthesize large-scale, diverse training data and incorporate domain randomization to enhance generalization. Evaluations in simulation and on a physical humanoid robot demonstrate that our method achieves high upper-body fidelity—mean joint-angle error < 8.5°—while guaranteeing zero falls during execution. It significantly outperforms baseline approaches in both motion accuracy and static/dynamic stability.

Technology Category

• Intelligent Robots: Manipulation
• Machine Learning: Imitation Learning & Inverse Reinforcement Learning
• Humans and AI: Human-Aware Planning and Behavior Prediction

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
• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems

Abstract

To support humanoid robots in performing manipulation tasks, it is essential to study stable standing while accommodating upper-body motions. However, the limited controllable range of humanoid robots in a standing position affects the stability of the entire body. Thus we introduce a reinforcement learning based framework for humanoid robots to imitate human upper-body motions while maintaining overall stability. Our approach begins with designing a retargeting network that generates a large-scale upper-body motion dataset for training the reinforcement learning (RL) policy, which enables the humanoid robot to track upper-body motion targets, employing domain randomization for enhanced robustness. To avoid exceeding the robot's execution capability and ensure safety and stability, we propose an Executable Motion Prior (EMP) module, which adjusts the input target movements based on the robot's current state. This adjustment improves standing stability while minimizing changes to motion amplitude. We evaluate our framework through simulation and real-world tests, demonstrating its practical applicability.

Problem

• Enabling humanoid robots to imitate human upper-body motions stably
• Addressing limited controllable range affecting standing stability
• Adjusting target movements to ensure safety and motion accuracy

Innovation

• Reinforcement learning framework for motion imitation
• Retargeting network generates large-scale motion dataset
• Executable Motion Prior adjusts target movements