Towards Professional Tennis Styles for Humanoid Robots with Adaptive Motion Planning and Tracking
Researchers introduced AdaPT, a hierarchical Adaptive Motion Planning and Tracking framework for humanoid robots. It extracts professional tennis serving and rally motions directly from broadcast video and applies them through separate planning and tracking layers.
The method targets the gap between stylistic quality and task accuracy in dynamic ball-handling scenarios. For automation and robotics coordination, it demonstrates a video-driven pipeline that separates high-level motion generation from low-level execution, allowing style constraints to be added without redesigning core controllers.
The framework's performance depends on the consistency and viewpoint diversity of available broadcast footage for training the motion models.