A Dynamic Control System for Robotic Goalkeeping Using YOLOv9 and Path Planning

This paper presents the design and implementation of a control system for a wheel-legged robot to function as a soccer goalkeeper. The proposed solution leverages a re-tuned, task-specific tiny YOLOv9 model to achieve a realtime balance between detection accuracy and speed. Using a depth camera, the system estimates the 3D position of a fastmoving soccer ball, while a two-phase path planning algorithm guides the robot to the optimal intercept point. To further enhance performance, the method mitigates noise interference during interception. Experimental results demonstrate the effectiveness of this approach, showcasing the synergy between deep learning techniques and classical control methods for dynamic and real-time robotic tasks.