The RoboCup agent behavior modeling challenge

Abstract

RoboCup is an international joint project that aims to foster Arti cial Intelligence (AI) and intelligent robotics research by providing a standard problem. RoboCup offers different challenges for intelligent agent researchers in a dynamic, real-time and multi-agent domain. One of these challenges, especially in the Simulation League, is the opponent modeling, which is crucial for the ultimate goal of the RoboCup project: develop a team of fully autonomous. In order to emphasize opponent-modeling approaches, the RoboCup Coach Competition was created and it was held every year (with some changes) from 2001 to 2006. Although there were several interesting research works about the agent modeling challenge during that time, several considerations were not well de ned and the competition was suspended after RoboCup Coach Competition 2006. In this paper, we propose a new approach for the competition to face the opponent modeling challenge in the RoboCup competition.