Procedural generation of balanced levels for a 3D paintball game

Videogames have become an important field in both the entertainment industry and the computational intelligence research community. Procedural Content Generation (PCG) allows to generate automatically interesting contents for a videogame with a low supervision from the game designers, or even without their supervision. This paper presents a search-based procedural content generator algorithm, that can create interesting maps in a 3D game using an evolutionary approach. Specifically, the algorithm is used in a tactical action game, based on the popular sport Paintball. The maps are generated with an objective in mind: they should be balanced, so any level should not provide an initial advantage of a team over the opponent. The algorithm proposed has been experimentally tested by generating different maps. These maps are represented by graphs of zones and populated with obstacles according to their densities. An evolutionary algorithm evolves these zones looking for an adequate distribution of obstacles in order to generate balanced maps. The experimental analysis shows how the algorithm is able to automatically create suitable maps for the game.