Search-based procedural content generation for race tracks in video games

Abstract

In recent years, Procedural Content Generation (PCG) has become a popular alternative for creating video games content. In this paper, we attempt to create a PCG for valid video game racing tracks while optimizing the fun they give to players. To achieve this, we represent a racing track as a sequence of parameterized segments, evaluate it using a multi objective fitness function, and then search for the optimum track by using search algorithms. Two search algorithms are used as alternatives to compare the effect it has on the racing tracks generated. The result we report show that the tracks generated can successfully match the validity criteria of being closed, continuous, and non-intersecting. More than that, they can also successfully create a certain degree of fun for players, according to data acquired from a questionnaire given to testers.