Generating Race Tracks With Repulsive Curves
Lasse Henrich, Falko Kötter · 2024
Procedural content generation is a useful tool for content creation in computer games. We present a novel approach for generating closed, arcade-like race tracks. Our system generates an initial shape by growing a curve inside a constrained space under self-repulsion, hence avoiding self-intersections while ensuring tight packing. This system is capable of generating a wide variety of closed race track shapes with deep customization options, like an isometric alignment of straight sections as developed here. Further, we devise algorithms for fitting the shape to a spline and introducing intersections, as well as fully generating a 3D model with features like crossings and bridges. We integrate the generated tracks into a game by training AI opponents and using them for automatic difficulty evaluation.