Design of intelligent strategies for simulated soccer player

Abstract

Simulated football is a useful tool for studying artificial intelligent strategies for multi-agent systems. In this project, a 2D football simulation platform called robocup is used as the developing environment. A simulation server program is provided as a football match environment controller. The client program is written by the user as the football player controller. The football match can be performed when the client program is connected to the server. This project introduces the design and implementation of the basic football player's framework. (1) Converting raw data from the sensor of the simulated football players into useful information, (2) Designing supporting functions to make use of the information collected to calculate and analyze the different situations of the soccer match condition and (3) Giving corresponding commands to the soccer players to perform different actions, like turn body, run, kick ball.....etc, based on the designed decision making functions and attacking strategies. In this paper, the designed framework of the players and the designed supporting functions will be presented. In addition, 1vs1 and 2vs1 attacking patterns will be demonstrated.