An evolutionary approach for strategy learning in RoboCup soccer

This paper proposes an evolutionary method for acquiring team strategies of RoboCup soccer agents. The action of an agent in a subarea is specified by a set of action rules. The antecedent part of action rules includes the position of the agent and the relation to the nearest opponent. The consequent part indicates the action the agent has to take when the antecedent part of the action rule is satisfied. The action of each agent is encoded by a integer string that represents the action rules. A chromosome is the concatenated string of integer strings for all the agents. The main genetic operator in our evolutionary method is mutation where a value of each bit is changed with a prespecified probability. Through computer simulations, we show the effectiveness of the proposed method as well as future research directions.