A multi-population genetic algorithm for procedural generation of levels for platform games

Abstract

This paper presents a multi-population genetic algorithm for procedural generation of levels for platform games such as Super Mario Bros (SMB). The algorithm evolves four aspects of the game during its generations: terrain, enemies, coins and blocks. Each aspect has its own codification, population and fitness function. At the end of the evolution, the best four aspects are combined to construct the level. The method has as input a vector of parameters to configure the characteristics of each aspect. Experiments were made to evaluate the capability of the method in generating interesting levels. Results showed the method can be controlled to generate different types of levels.