Abstract: Focused on the problem of modeling matching between ideal systems and virtual systems in the research on robot sheepdog, during the design of robot driving controller, fuzzy control theory and genetic algorithm are syncretized and optimized to improve robot moving precision. Finally a harmonious control of robot two drive motors is achieved. In this paper, GA operators are optimized, and a new method is proposed to control many complex systems whose models are unknown, using this method fuzzy rules can be acquired automatically according to training samples. The simulation result shows the effectiveness and feasibility of the method.