Playing Soccer Like Humans Do: Learning Versatile and Agile Humanoid Soccer via Latent Motion Priors
Playing Soccer Like Humans Do: Learning Versatile and Agile Humanoid Soccer via Latent Motion Priors
This research presents soccerbot, a hierarchical reinforcement learning framework that combines agile locomotion and ball manipulation for humanoids. By using latent motion priors distilled from large-scale human motion capture data, the system enables unified skill execution including moving, dribbling, trapping, and kicking with human-like motion quality. The approach transfers zero-shot to the Unitree G1 humanoid robot, demonstrating robust performance in real-world conditions.
The paper introduces soccerbot, a hierarchical reinforcement learning framework that uses latent motion priors distilled from human motion capture data to enable a humanoid robot to perform agile locomotion and ball manipulation tasks like dribbling and kicking with natural transitions.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- paper: not_reported
- code: not_reported
- weights: not_reported
- dataset: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
unified formulation, while inherently producing human-like motion and fluid transitions during commanded skill switching. Augmented by robustness-oriented simulation design, soccerbot transfers zero-shot to the Unitree G1 humanoid robot, demonstrating agile, natural, and robust execution of the full soccer skill repertoire under real-world conditions.
Open source S4
tion quality unaddressed. We present soccerbot, a hierarchical reinforcement learning framework that resolves both limitations through a structured latent motion prior distilled from large-scale human soccer motion capture data. By constraining policy learning within this dynamically plausible motion manifold, a single skill-conditioned policy naturally learns to perform moving, dribbling, trapping, and kicking in a
Open source S3
arXiv:2611.06564v1 Announce Type: new Abstract: Humanoid soccer requires a robot to seamlessly combine agile locomotion, ball manipulation, and impact-rich contact dynamics across a continuous, dynamic behavioral spectrum. Yet existing approaches either cover only a limited set of skills, or forego motion priors entirely and produce unnatural behaviors, leaving the combination of a unified skill set and human-like mo
Open source S2
Source：arXiv Robotics — research abstracts · arxiv.org