Skip to main content

Projects

Current projects

Scheherazade
Using machine learning and crowdsourcing to create an "open narrative intelligence" that creates and understands stories about almost any conceivable topic.
[Pubs]
Quixote: Learning from Stories
A robot and virtual agent architecture for translating natural language stories into reinforcement learning behavior. Learning from stories is a technique for using natural language stories as demonstrations.
[Pubs]
Learning Game and Level Generation from YouTube Videos
Using computer vision to learn and use game design knowledge from YouTube videos. Demonstrated in the Super Mario Bros. game.
[Pubs]
Procedural Content Generation for Augmented Reality Games
Procedural content generation algorithms can be used to make augmented reality games more interesting.
[Pubs]
Understanding Mechanics in Games with a Purpose
Games with a Purpose are games in which playing results in solving a serious problem or generating a data set. Little is known about what game mechanics yield improved results and player performance. This project explores different mechanics and how the affect data quality and player enjoyment in games that collect commonsense knowledge.
[Pubs]
Game Mechanic Generation
Using machine learning to generate complete games from scratch.
[Pubs]
The Lovelace 2.0 Test of Computational Creativity
An alternative to the Turing Test that can measure an intelligent system's creative ability.
[Pubs]


Older projects

Optimizing Player Experiences
Data-driven personalized drama management using sequential collaborative filtering.
[Pubs]
CAPTCHA games
A computer game alternative to ReCAPTCHA. CAPTCHA games can be dynamically generated.
[Pubs]
GameForge
A procedural computer role playing game generator.
[Pubs]
Dramatis: Computational Model of Suspense
A computational model of the affective suspense response to narrative content.
[Pubs]
Proactive Sensing
Proactive sensing games are games that, when played, collect information about the current state of the real world that is otherwise hard to with other, passive sensors such as cameras. We demonstrate proactive sensing in a prototype game that collects the locations of handicap ramps and other accessibility features of the real world.
[Pubs]
GameTailor
Data-driven procedural content generation using collaborative filtering and declarative optimization.
[Pubs]
Assistive AI in Machinima End-User Creation
How can intelligent support tools can assist human creators of machinima? (with DWIG).
[Pubs]
Digital Improv and Computational Play
Seeks to understand the creative processes involved in improvisational theater (with Adam Lab).
[Pubs]
Reflex/React
Social skills training intervention for children with high-functioning autism spectrum disorders.
[Pubs]
Doraemon Generator
Generates gadgets for sci-fi & fantasy stories using conceptual blending. Inspired by the Doraemon comics.
[Pubs]
WeQuest: Mobile Alternate Reality Games
Alternate reality games overlay fictional elements on the real world. WeQuest is a system for running mobile, GPS-enabled ARGs that uses artificial intelligence to "translate" playable stories from one city to another.
[Pubs]
Scenario and Plot Adaptation via Planning
Automatically "rewrite" training scenarios and game plots to specification.
[Pubs]
Story Generation with Case-based and Analogical Reasoning
Case-based & analogical reasoning as a model for creativity in story generation.
[Pubs]
Fabulist Story Generator
A story generation architecture based on cognitive principles of narrative intelligence.
[Pubs]
Cambot and Comicbot
Cambot is a virtual move director for 3D worlds that produces machinima visualizations from an annotated script. A modified version, Comicbot, creates sequences of comic-book like stills with text bubbles for dialogue.
[Pubs]
Automated Story Director
A framework for creating AI-controlled interactive narratives for entertainment or training.
[Pubs]





[Miscellaneous publications] include papers not directly associated with the above projects.