Dribble Master: Learning Agile Humanoid Dribbling Through Legged Locomotion

2025-05-19

AI Summary

Humanoid robot dribbling in soccer requires simultaneous high-precision ball control and dynamic balance maintenance; conventional rule-based approaches are limited by fixed gaits and poor real-time adaptability. This paper proposes a two-stage curriculum reinforcement learning framework that end-to-end learns a fully autonomous policy—from basic locomotion to fine-grained dribbling—without predefined trajectories or explicit dynamical models. We innovatively introduce a virtual camera model to enhance active visual perception, integrate heuristic reward shaping, and employ a Sim2Real transfer mechanism, enabling the first vision-guided, human-like agile dribbling on physical humanoid platforms. The method is successfully deployed across diverse real-world terrains, achieving stable, smooth, and omnidirectional dribbling with natural motion quality and significantly improved environmental robustness over baseline methods.

Technology Category

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

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Humanoid soccer dribbling is a highly challenging task that demands dexterous ball manipulation while maintaining dynamic balance. Traditional rule-based methods often struggle to achieve accurate ball control due to their reliance on fixed walking patterns and limited adaptability to real-time ball dynamics. To address these challenges, we propose a two-stage curriculum learning framework that enables a humanoid robot to acquire dribbling skills without explicit dynamics or predefined trajectories. In the first stage, the robot learns basic locomotion skills; in the second stage, we fine-tune the policy for agile dribbling maneuvers. We further introduce a virtual camera model in simulation and design heuristic rewards to encourage active sensing, promoting a broader visual range for continuous ball perception. The policy is trained in simulation and successfully transferred to a physical humanoid robot. Experimental results demonstrate that our method enables effective ball manipulation, achieving flexible and visually appealing dribbling behaviors across multiple environments. This work highlights the potential of reinforcement learning in developing agile humanoid soccer robots. Additional details, video demonstrations, and code are available at https://zhuoheng0910.github.io/dribble-master/.

Problem

• Enabling humanoid robots to perform agile soccer dribbling
• Overcoming limitations of rule-based ball control methods
• Transferring simulated dribbling skills to physical robots

Innovation

• Two-stage curriculum learning for dribbling
• Virtual camera model for active sensing
• Heuristic rewards for continuous ball perception