Agility Meets Stability: Versatile Humanoid Control with Heterogeneous Data

2025-11-21

AI Summary

Addressing the challenge of simultaneously achieving agile dynamic motion tracking and extreme balance maintenance in humanoid robots, this paper proposes the first unified end-to-end control framework. Methodologically, it integrates human motion-capture data with synthetically generated balance data, employs a hybrid reward mechanism, task-specific reward shaping, and performance-driven heterogeneous sampling, and incorporates physics-constrained modeling with deep reinforcement learning. The core contribution is the first integration of dynamic trajectory tracking and extreme balance capabilities within a single policy—enabling zero-shot transfer of balance skills across tasks. Evaluated in simulation and on a real Unitree G1 robot, the framework achieves high-fidelity execution of agile behaviors—including dancing and running—as well as extreme balance maneuvers such as the “Ip Man squat,” demonstrating substantial improvements in generalization and robustness.

Technology Category

• Intelligent Robots: Manipulation
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Computer Vision: Motion & Tracking

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Humanoid robots are envisioned to perform a wide range of tasks in human-centered environments, requiring controllers that combine agility with robust balance. Recent advances in locomotion and whole-body tracking have enabled impressive progress in either agile dynamic skills or stability-critical behaviors, but existing methods remain specialized, focusing on one capability while compromising the other. In this work, we introduce AMS (Agility Meets Stability), the first framework that unifies both dynamic motion tracking and extreme balance maintenance in a single policy. Our key insight is to leverage heterogeneous data sources: human motion capture datasets that provide rich, agile behaviors, and physically constrained synthetic balance motions that capture stability configurations. To reconcile the divergent optimization goals of agility and stability, we design a hybrid reward scheme that applies general tracking objectives across all data while injecting balance-specific priors only into synthetic motions. Further, an adaptive learning strategy with performance-driven sampling and motion-specific reward shaping enables efficient training across diverse motion distributions. We validate AMS extensively in simulation and on a real Unitree G1 humanoid. Experiments demonstrate that a single policy can execute agile skills such as dancing and running, while also performing zero-shot extreme balance motions like Ip Man's Squat, highlighting AMS as a versatile control paradigm for future humanoid applications.

Problem

• Unifying agile motion tracking and robust balance maintenance in humanoid robots
• Reconciling divergent optimization goals between agility and stability
• Developing versatile control policies using heterogeneous motion data sources

Innovation

• Unified policy for dynamic motion and balance
• Hybrid reward scheme with heterogeneous data sources
• Adaptive learning strategy for diverse motion training