Procedural generation of rollercoasters

The "RollerCoaster Tycoon" video game involves creating rollercoasters that optimize for various in-game metrics, while also being constrained by the need to ensure a feasible structure in terms of physical and spatial bounds. Creating these procedurally is thus a challenge. In this work, we explore multiple approaches to rollercoaster generation, including Markov chains and machine learning and reinforcement learning algorithms. We show that we can achieve relatively good tracks in terms of the game’s measurement of success, and that reinforcement learning may give more control over other factors of potential interest. A focus on multiple measures allows our work to extend to other factors that also mimic actual player constructions.