Bachelor's Degree Thesis
Automated playtesting of dungeons using Goal-Oriented Action Planning.
Different personas rated the priority of actions differently.
Made in Unity (C#).
Bachelor's Degree in Computer Science.
For my Bachelor's Degree, me and a fellow student developed AI-agents with different personas (Rusher, Monster Slayer & Treasure Hunter) that were able to solve procedually generated dungeons (2D) using a Goal-Oriented Action Planning - algorithm. We developed everything in Unity and the maps were created in The Evolutionary Dungeon Designer ("EDD").
We tested the personas on a set of different maps and collected data on how fast different personas could solve different maps. Some personas "enjoyed" different types of maps and some other types. This data could then potentially be used by level designers to evaluate the difficulty and enjoyment of their creations or dynamically alter maps in realtime to suit the current player.
The thesis paper can be sent upon request!