Design and Implementation of ROS-Based Intelligent Ball Picking Robot

Ball sports are accompanied by a significant amount of ball picking tasks. Currently, intelligent ball-serving robots lack the capability for ball retrieval, making manual collection inefficient and adding to the athletes' training burden. This paper, based on the ROS system, integrates laser lidar, IMU sensors, realsense D435i cameras, YOLO object detection algorithms, visual SLAM, gmapping graphic optimization, and path planning algorithm to implement a type of ball-picking mobile robot. The robot by using deep learning algorithms, the number and position of balls are calculated and identified precisely. The designed and developed ball-picking robot called Probot, which demonstrates strong adaptability to various environments. Experiments show that Probot able to accurately position and map in dynamic environments, autonomously plan paths, and avoid obstacles. The application of this technology is not limited to ball-picking robots, it can be applied to other robots and automated devices to enhance their performance and efficiency.