Procedural Level Design Using an Interactive Cellular Automata Genetic Algorithm

In this paper we propose an interactive genetic algorithm to evolve maze levels for computer games. We represent a maze with a cellular automaton and the genetic algorithm evolves the cellular automata rules applied to a starting maze level state. Users then rate the fitness of a subset of the generated population using an image of the top-down view of the maze. After ten generations, users then play through the best evolved maze within a maze runner type game using a first person perspective. User ratings show that our IGA was able to evolve highly rated mazes.