Educational Use of "Game AI Competition Portal" for Advanced Programming Courses

It is not surprising that competitions have been important tools to promote research and education in challenging problems. Recently, there have been a lot of game AI (Artificial Intelligence) competitions for video games, real-time strategy games and first person shooting games. The purpose of the game AI competitions is to build a program to play games automatically without human intervention. Because materials for competitions (rules, manuals, software, samples and so on) have been open to the public, it is promising to use them for education. Educators can adopt them for their courses' project by focusing only on the AI issues. It has been known that game AI competitions have been a good benchmarking tool and it enable students compare their works with state-of-the arts. However, it's not easy to get basic information on the game AI competitions if the instructor is not familiar with the competitions. In this paper, we propose to use Game AI competition portal (http://cilab.sejong.ac.kr/gc) with the history, results, AI techniques, source code, recent and future competitions and links for several game AI competition for education purposes. Introduction Recently, there have been a lot of game AI competitions in the world ranging from board games (Othello and Go) to complex real-world games (car racing, StarCraft, unreal tournament, and so on). The purpose of the Game AI Competitions is to build a program to play games automatically. However, it is not trivial to start the game AI competition for beginners because the contents are designed for experts. It is necessary to provide meta-level portal of the game AI competitions with easy introduction and step-by-step guidelines. Usually, it requires much effort to design programming projects for advanced computer science courses. Educators should define problems (sometimes, open-ended) to be solved by students. If the problem set is invented by the educator for the first time, it is not easy to compare student's works with current state-ofthe art. In addition to that, the introduction of multimedia and games (may be interesting to students) mean extra skills to edit/modify the programs for the course projects. We have introduced the use of game AI competitions for education [1] with companion game AI competition portals. Useful information on several famous game AI competitions has been posted in this portal. It includes the history of previous competitions, rules, results, winning strategies, source codes and educational materials. This game AI competition portal 1 http://cilab.sejong.ac.kr/gc * Corresponding Author can be a good starting point for those who have interested in game AI competitions. Game AI Competitions Games have been widely used as a test bed for artificial intelligence and computational intelligence research. It gives researchers concrete goals and ways to evaluate the goodness of their approach. In the game AI competitions, organizers define concrete goals and operational rules of the competition, and provide with free software, API, example controllers, and documents. The task of each participant is to develop their own AI controller for the game and submit it to organizer when it is ready. This approach is quite successful and efficient because each research team can skip the development of the details of goals, rules, platforms and comparisons with other methods. The organizers run the competition in the conference day and determine the winner of the competition. For several years, game AI competitions have been important parts of several conferences and journals. Several researchers invested their invaluable time and effort on the competition by organizing the events, managing websites, making competition software and documents. It resulted in several international conference and journal papers related to the competitions and media appearance through major magazines and YouTube. It has opened several new research areas: Learning from human behaviors, Turing test for game AI, Procedural contents generation, and User Satisfaction Modeling. The competition evolved 2014년 한국컴퓨터종합학술대회 논문집