An Application of ASP for Procedural Content Generation in Video Games

Procedural content generation eases and accelerates the development of video games by creating data algorithmically through a combination of human-generated assets and algorithms usually coupled with computer-generated randomness. This paper presents a use case of Answer Set Programming (ASP) for procedural content generation of levels in a rougelike video game powered by the Godot Engine. The main elements of a set of human-generated rooms are represented by ASP facts, among them positions of doors, presence of treasures and power-ups. Within this knowledge, ASP is asked to generate dungeons satisfying a few conditions, among them the correct positioning of rooms, the absence of unreachable rooms and constraints on the occurrences of rooms. Scalability of ASP in this context is evaluated empirically, showing that it can generate in few seconds levels that comprise thousands of rooms.