Humanoid Robot Soccer
Motion learning, action retargeting, and closed-loop control for RoboCup 3D soccer.
Overview
This work supported a RoboCup China 3D Simulation League team that received the National First Prize and finished as runner-up in 2026. The project focuses on reliable humanoid actions and coordinated 7-vs-7 robot soccer.
Contributions
The system retargets AMASS human motion to the Booster T1 robot and trains policies to track kick and locomotion trajectories. It also uses fine pose alignment and proportional control to improve balance, contact accuracy, and action transitions during play.