Game Forge


Role-playing games, and other contemporary video games, usually contain a main storyline consisting of several causally related quests. As Game players usually possess diverse motivation, tastes, preferences, it can be beneficial to customize game quests. As the cost of hiring enough human labor to cater to each individual player is prohibitively expensive, we propose an AI can adapt human-authored plotlines to generate customized game narratives. Our system facilitates the authoring of game quest and gives designers a “leverage” with which they can save substantial efforts.

Quest-centric game adaptation is capable of (1) generating a large variety of quest combinations to suit the need of each individual player and enhance re-playability, (2) maintaining story quality by minimizing irrelevant events and wasted effort during the game and (3) balancing the preservation of the original stories and the adaptation in order to exploit and facilitate human creativity.

We employ a partial-order planning approach to the quest-centric game adaptation problem. The planner takes a complete storyline comprised of several quests and iteratively searches for modifications, deleting and inserting quests and events, until it meets the user’s preferences. Stories generated conform to a formalized notion of narrative coherence, which improves the aesthetic appeal of the stories.

The quest is used to generate and execute a Zelda-style interactive game world. The game world is generated using a genetic algorithm that searches for a world that services the quest and also achieves player preferences.