Towards a generic framework for automated video game level creation

Abstract

This paper presents a generative system for the automatic creation of video game levels. Our approach is novel in that it allows high-level design goals to be expressed in a top-down manner, while existing bottom-up techniques do not. We use the FI-2Pop genetic algorithm as a natural way to express both constraints and optimization goals for potential level designs. We develop a genetic encoding technique specific to level design, which proves to be extremely flexible. Example levels are generated for two different genres of game, demonstrating the system's broad applicability. © 2010 Springer-Verlag Berlin Heidelberg.