Genetic Programming based Optimisation of Soccer Agents

This paper describes our approach on evolving a robot controller to generate a set of goalkeeper behaviours in the RoboCup domain based on the genetic programming (GP) technique. The goalkeeper agent is a software construct that operates in a simulated environment provided by the Soccer Server. Our research is to determine the conditions that need to be met for an evolved goalkeeper agent to perform adequately in a defensive situation. A framework is developed to test the agent’s performance in the simulator. The impact of the sensors information that operates as the input to our GP implementation is addressed. The framework by incorporating it within two established soccer agents is evaluated, and then the offline performance of each agent is tested. The experimental results show that our approach is able to produce a robust robot controller and the framework is flexible enough to be readily incorporated into other existing soccer agent frameworks.