Intelligent Maze Generation Based on Topological Constraints

Maze is a puzzle, in which players need to find a path from an entrance to a goal on the maze. Although maze is a puzzle itself, it is also used as a tool in different fields, such as computer game and robotics. Since maze users in different fields may have different purposes of using a maze, different desired properties may be wanted on the maze. This paper provides a method which generates the desired maze when users give desired properties. Our method uses search-based procedural content generation (SBPCG) approach, in which the process repeatedly generate-and-test mazes to obtain a satisfactory maze. Our research focuses on a perfect maze, and existing perfect maze generation algorithms can be used in the SBPCG approach. When the algorithms are used, to obtain a desired maze with a higher probability, our method chooses the best algorithm amongst intelligently. Lastly, we provide several use cases and demonstrate that our method generates desired mazes effectively for each use case.