Evaluation of Team Strategies Using Match Results for an Evolutionary Computation Method

Abstract

In this paper, an evaluation method is proposed to improve the performance of an evolutionary computation method for RoboCup soccer team strategies. The averaged scores are used in the proposed method to handle uncertainty in the game. Competent team strategies with high averaged scores and low conceded scores are used to generate next team strategies in the evolutionary computation. Thorough computer simulations we show the effectiveness of the proposed method.