An Evolutionary Algorithm for Procedural Content Generation for the game Super Mario Bros

Abstract

Procedural content generation (PCG) is a method of automated creation of digital content. One of the challenges in the area is to generate a variety of content, with low computational cost and guaranteed validity. The present paper addresses the procedural generation of typical levels of Super Mario Bros (SMB) game. This project considers the representation of a SMB level, by using a vector of vertical slices; accordingly, it is possible to create a set of patterns and use them in the generation of the map. Therefore, the aim is to use Evolutionary Algorithms (AEs) combined with the patterns (set of slices from a real game phase) to generate the SMB levels. The preliminary results showed that, despite the use of few standards, the conducted experiments converged to feasible levels.