Design of Intelligent Strategies for Simulated Soccer Players

Abstract

Simulated soccer is an excellent domain for testing artificial intelligent strategies for multi-agent systems. The goal of this project is to design intelligent strategies for controlling a team of virtual soccer players. The main area focused is to develop an effective offensive attack strategy so as to kick the ball into opponent's goal. The strategy is implemented and evaluated on a simulation platform called Tao of Soccer, which is a java-based simulation platform. The platform creates 2D soccer field, the players and the ball. It also enforces the rules of the soccer game. The soccer players in the field have the same physical condition such as speed, maximum kick force, acceleration etc. so that the development only focuses on the strategy. The evaluation of the strategy is based on the number of goals scored over a limited time and the opponent is the default simple AI provided by the platform.