A Real-time Football Goalkeeper Robot System Based on Fuzzy Logic Control

Abstract

This paper develops a football goalkeeper robot system that can entertain people or assist routine training in indoor or outdoor environments. The system uses an Intel RealSense D435 depth camera to provide image data and the yolov3-tiny Net to locate the football. In the process of coordinate transformation, the least square method was utilised to obtain the seven parameters, so that the error can be kept within an acceptable range (within the pre-specified range).Using the fuzzy logic control algorithm to predict the hitting points on the football baffle and PID algorithm to control motor rotation to prevent the football from entering crossbar. In the experiments, comparison between the fuzzy logic control algorithm and the Kalman filter algorithm was implemented. The results shows that the fuzzy logic control has a better performance.