Perception and decision optimization of basketball robot based on YOLOv5 and LiDAR fusion

A basketball intelligent robot often faces interference in target detection, limited tracking stability, and insufficient multisensor fusion in dynamic and complex environments. This study proposes an optimized basketball intelligent robot system that integrates YOLOv5 with Light Detection and Ranging technology. The system builds a multimodal perception and control framework based on data collaboration. It combines multi-target tracking and trajectory prediction mechanisms to improve the intelligent decision-making ability of the basketball robot. Performance testing indicates the detection accuracy is 97.85%, and the tracking loss rate is 6.12%. Taking autonomous interception and cooperative defense as tasks, the success rate is 94.28 and 92.56%, respectively. In particular, the experimental results demonstrate that across various basketball scenarios, such as rapid dribbling or multi-person occlusions, the tracking success rate remains above 97.13%. The experimental results indicate that the optimized solution is highly effective, distinguishing it significantly from previous methods. The proposal from this study is a novel idea for designing intelligent sports robots.