Model AI Assignments

The Model AI Assignments session seeks to gather and disseminate the best assignment designs of the Artificial Intelligence (AI) Education community. Recognizing that assignments form the core of student learning experience, we here present abstracts of eight AI assignments that are easily adoptable, playfully engaging, and flexible for a variety of instructor needs. Assignment specifications and supporting resources may be found at http://modelai.gettysburg.edu. The Pac-Man Projects Software Package for Introductory Artificial Intelligence John DeNero, Dan Klein The Pac-Man projects apply a variety of artificial intelligence (AI) techniques in the setting of the classic video game Pac-Man. Despite their video game theme, the projects are targeted very broadly. They cover a range of foundational AI concepts, including informed state-space search, probabilistic inference, and reinforcement learning. These concepts underlie real-world application areas such as natural language processing, computer vision, and robotics. The Pac-Man projects promote effective learning through several design principles. Graphical interfaces allow students to visualize the results of the techniques they implement. Harness code contains commented examples and clear directions, but does not force students to wade through undue amounts of scaffolding. Finally, the domain of Pac-Man itself provides a challenging problem environment that demands creative solutions; real-world AI problems are challenging, and Pac-Man is too. In our course at UC Berkeley (CS 188), these projects have substantially boosted enrollment, teaching reviews, and student engagement. The projects have been field-tested, refined, and debugged over multiple semesters. We are now excited to release them to Copyright c © 2010, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved. other universities for instructional use. This software package includes four primary projects on search, multi-agent search, reinforcement learning, and probabilistic tracking. It also includes an open-ended final contest and an initial Python tutorial. A Project on Fast Trajectory Replanning for Computer Games for “Introduction to Artificial Intelligence” Classes Sven Koenig, William Yeoh This standalone path-planning project for an undergraduate or graduate artificial intelligence class relates to video game technologies and is part of our effort to use video games as a motivator in projects without the students having to use game engines. Heuristic search and, in particular, A* are among the most important search techniques and thus are good candidates for a project in artificial intelligence. In this project, the students need to code A* and then extend it to Adaptive A*, a fast trajectory replanning algorithm, to move game characters in initially unknown gridworlds to a given target. Adaptive A* is an incremental version of A* that often searches faster than A* since it updates the heuristics between searches to find solutions to series of similar search tasks potentially faster than is possible by solving each search task from scratch. This project requires students to develop a deep understanding of A* and heuristics to answer questions that are not yet covered in textbooks. The project is versatile since it allows for theoretical and implementation questions. We list a variety of possible project choices, including easy and difficult questions. The project text and additional support material (such as a maze generator and pointers to the literature) can be found at http://idm-lab.org/gameai. Getting Set with OpenCV Zachary Dodds “Getting Set with OpenCV” is a two-week assignment used as a hands-on introduction to computer vision in an undergraduate AI course. Set is a game of visual percep1919 Proceedings of the Twenty-Fourth AAAI Conference on Artificial Intelligence (AAAI-10)