On using evolutionary computation approach for strategy optimization in robot soccer

This paper proposes an approach for design and optimization of strategy in the domain of robot soccer. The strategic placements, evolved via evolutionary algorithms, allow the team to enhance their decision-making capability for real-time strategy execution. Experiments are conducted using binaries of teams participating in RoboCup Soccer Simulation 3D League competitions. In the context of robot soccer, strategy denotes the critical areas where home team agents should be positioned keeping in view the strength and weakness of the opponent agents. For simplicity, only defensive scenarios are considered. Thus, the goal difference, regions occupied by defending players and ball possession are used as the fitness criteria. The results of the experiments clearly demonstrate that strategy optimization against a specific opponent contributes radically towards a team's success.