3D vision based fast badminton localization with prediction and error elimination for badminton robot

摘要

In this paper, the problem of fast badminton localization problem is investigated for a class of badminton robots. More precisely, a manifold-learning based localization method is implemented for the improvement of hitting accuracy and effectiveness. Based on the localization results, a novel badminton trajectory prediction algorithm is designed based on 3D Vision in the real world. Furthermore, clock-synchronization combined with motion compensation methods are also proposed to better localization error elimination. In the end, the validity and usefulness of our proposed algorithm is demonstrated by numerical experiments.