A Fuzzy Mechanism for Action Selection of Soccer Robots

A vision-based soccer robot system is proposed, in which vision will identify the position and heading angle of each robot, and the position of the ball. With these imaging data, values for the defense factor, the competition factor, and the angle factor, are obtained. Using the three factors as the input variables of the proposed action selection mechanism, which is expressed in terms of fuzzy rules, the action that each robot should take is then chosen from the five basic actions available for the robots. In this manner, each robot may intercept, shoot, block, sweep the ball, or just stand by. After determining the action of each robot, control commands generated by the host computer are sent to the robot directly through a wireless RS-232. To show the feasibility of the proposed method, experimental results of a robot soccer game will be used for illustration.