A project used to write a Bachelor’s Thesis in Computer Science at Malmö University.
abstract:
This paper aims to explore an alternate way about how game testing of procedurally generated dungeons in role playing games can be conducted automatically with archetypical player personas. For this purpose, the Automatic Dungeon Evaluator (ADE) was developed featuring an AI-agent. In its decision-making process, the agent uses Goal Oriented Action Planning (GOAP) to traverse a dungeon and produce a rating depending on that agent’s current persona. This rating can be used by developers and designers of games or researchers alike in order to improve or explore patterns in rooms or entire dungeons and how well this suits different play-styles.