Opponent Modeling in RoboCup Soccer Simulation

摘要

RoboCup is an international scientific initiative with the goal to advance the state of the art of intelligent robots. RoboCup offers an integrated research task covering broad areas of Artificial Intelligence and robotics. Within this competition and without the necessity to maintain any robot hardware, the RoboCup Simulation League is focused on artificial intelligence and team strategy. This league can be considered as a multi-agent domain with adversarial and cooperative agents where the team agents should be adaptive to the current environment and opponent. In this paper, we present an approach for creating and recognizing automatically the behavior of a simulated soccer team.