Automatic Ball Launcher Machine for Goalkeeper Training Using YOLO Deep Learning Algorithm

Soccer is one of the most popular sports in the world, and goalkeepers play a crucial role in the sport. However, in their training, many soccer clubs still rely on manual methods and require teammates for practice. Therefore, there is a need for a ball launcher device. The existing ball launcher devices are limited to machines that can throw the ball in a desired direction. On the other hand, research on fast object detection using the YOLO (You Only Look Once) algorithm is rapidly advancing. However, its implementation in the field of sports is still limited. There is a need for collaboration between YOLO algorithm research in the field of soccer to create a ball launcher robot. In the development of this automated ball launcher robot, it can detect the goalkeeper and goal objects using the YOLO algorithm. The average confidence score in this study is 80.56% for 5 goalkeepers, and the confidence score for the goal is 85.46%. The goal is divided into 3 target positions: right, left, and center. Based on the lookup table results, the RPM value for position A is 2335 RPM for motor 1 and 463 RPM for motor 2. Position B is 1469 RPM for motor 1 and 1431 RPM for motor 2. Position C is 472 RPM for motor 1 and 2151 RPM for motor 2. Based on the results obtained in this study, it is hoped that this device can be beneficial for goalkeeper training.