Intelligent table tennis recycling approach based on improved Siamese network tracking

With the popularization of intelligent auxiliary equipment in sports training, the efficient recycling of table tennis balls has become a key factor in improving training efficiency. The motivation of this study is to solve the problem of robotic arm grasping failures caused by visual perception lag in traditional recycling equipment. A smart table tennis recycling method oriented towards practical engineering applications is proposed. This study pioneers the end-to-end integration of an improved Siamese network target tracking framework with a robotic arm hardware system. This method combines template update strategy and channel attention mechanism, and carries out small-target parameter tuning and positioning error compensation at the system level, thereby effectively coping with the challenges of target tracking and robotic arm collaborative grasping for table tennis in high-speed motion, complex backgrounds, and occluded environments. In clear scenes, the EAO reaches 0.85 and the SR is 93.1%. An EAO as high as 0.85 effectively ensures stable tracking during the recycling process, significantly reducing the misoperation rate of the mechanical arm, and the end-to-end recycling time of the system is only 1.9 s.