Co-evolving Soccer Softbot team coordination with genetic programming

Abstract

In this paper we explain how we applied genetic programming tobehavior-based teamcoordination in the RoboCup Soccer Server domain.Geneticprogramming is a promising new method for automatically generatingfunctionsand algorithms through natural selection. In contrast to other learningmethods,genetic programming’s automatic programming makes it a natural approachfordeveloping algorithmic robot behaviors. The RoboCup Soccer Serverwas a verychallenging domain for genetic programming, but we were pleased withtheresults. At the end, genetic programming had produced teams of soccersoftbotswhich had learned to cooperate to play a good game of simulator soccer.