GENERATING EMERGENT TEAM STRATEGIES IN FOOTBALL SIMULATION VIDEOGAMES VIA GENETIC ALGORITHMS

This paper defends the use of evolutionary algorithms to generate (and evolve) strategies that manage the behavior of a team in simulated football videogames. The chosen framework to develop the experiments is Robocup, an internatonal project that promotes the use of Artificial Intelligence in socially significant areas. One of these areas is related to computer games in the form of a simulated soccer league that allows two teams of 11 simulated robotic autonomous players to play football without human intervention. This paper proposes to generate emergent behaviors for the teams, via an evolutionary training process. The proposal is an alternative to implementing specific AI controllers for both players and teams in football videogames