Research on Goalkeeper Strategy Based on Random Forests Algorithm in Robot Soccer

In nowadays, most of the goalkeeper´s strategies in robot soccer teams are on the basis of prediction of the ball motion trajectory. In order to improve goalkeeper´s intelligence in robot soccer match, this paper presented a robot soccer goalkeeper´s strategy based on random forests algorithm. This strategy predicted the movement trends of the opposite robots and ball near our forbidden zone by use of random forests algorithm; According to information of predicted location, threat degree analysis model is applied to analyze the threat degree of opposite robots, and the goalkeeper is able to make correct defense with the threat degree. Simulation experiments showed the validity of this strategy.