Humanoids learn soccer skills using depth images and joint control
DAVIS: A Depth-Only End-to-End Active-Vision Framework for Humanoid Soccer Skills
Robotics
Summary
Playing soccer with humanoid robots is tricky because they need to see the ball, move their legs just right, and adjust to changing views while playing. The authors created DAVIS, a system that teaches robots to use only depth images from a camera on their head, plus their own body information, to control all their joints for soccer moves like shooting and dribbling. This system learns by practicing in simulations and real robots, without needing extra steps to plan or recognize objects during play. Their approach helps robots keep track of the ball and keep playing even when views change or contacts with the ball are uncertain.
What this means in practice
- •For humanoid robot developers: Enable humanoid robots to perform soccer skills with depth-only perception and joint control without separate planning modules.
- •For robotic manipulation engineers: Inspire manipulation tasks where robots learn contact skills using limited sensing and direct joint control from raw depth data.
Authors
Jiakang Jin, Yixiao Huo, Pengyuan Wang, Yinan Han, Tingxuan Zhang, Zhuobing Zhao, Xuanxin Zhou, Zhangchen Ye, Enxuan Ruan, Yifei Bao, Jiankun Yang, Chenghao Sun, Wenhao Cui, Xiaoyu Tian, Yiming Li
Abstract
Humanoid soccer contact skills require more than producing high-impact foot-ball contacts: the robot must close the loop over perception, approach, alignment, impact, and recovery while its own motion induces substantial viewpoint changes, frequent loss of the ball from view, and uncertain contact outcomes. In this work, we ask a compact yet stricter question: can a humanoid learn soccer contact skills using only a head-mounted depth image, proprioceptive history, and an optional low-dimensional task command, and directly output 25-DoF joint PD targets without extra runtime perception or planning modules? To this end, we propose DAVIS, a depth-only end-to-end framework for humanoid soccer skills that learns visibility-aware auxiliary geometry during training, and combines GT-to-prediction annealing, task curricula, and AMP-style motion priors to smoothly bridge privileged supervision and real deployment. Built on this framework, we instantiate representative soccer contact skills, including goal-directed shooting and directional dribbling, through task-specific definitions of objects, commands, rewards, and curricula, and validate them through simulation, Noetix E1 real-robot experiments, and ablations.