Video Games and Artificial Intelligence
Adi Botea, Ralf Herbrich and Thore Graepel
Artificial Intelligence (AI) and video games have traditionally been connected through the use of AI to control non-player characters (NPCs). From a research perspective, they are emerging as a great mutually beneficial combination. On the one hand, AI technology can provide solutions to an increasing demand for realistic, intelligent behaviour of NPCs. On the other hand, as games become more complex and realistic simulations, they offer a range of excellent testbeds for AI research.
This tutorial gives an introduction to the application of AI techniques, such as learning, search and planning, to video games. The tutorial focuses on methods, applications, open problems, and resourcesavailable to people who want to work in this space. Concrete examples of AI challenges covered in this tutorial include driving a car in a racing game, player rating and matchmaking, path finding on a map, and planning the behaviour of NPCs. Several demonstrations are available and make the tutorial concrete, vivid and entertaining.
The tutorial is intended for both researchers looking for realistic benchmarks to test new algorithms and ideas, and for game developers looking for ways to improve their products. No deep prior knowledge is required in either of the covered topics.
Adi Botea has obtained a PhD degree at the University of Alberta. He holds a research and lecturing position at NICTA and the Australian National University. The main interest areas are planning, search and games. Games-related work includes hierarchical pathfinding, multi-agent pathfinding, planning in video games, and generating crosswords grids.
Ralf Herbrich obtained his PhD in statistics from the Technical University of Berlin before joining Microsoft Research Cambridge, where he is now heading the Applied Games/OSA group together with Thore Graepel. His research interests include machine learning, online gaming and ranking systems, probabilistic modeling, programming languages and statistical learning theory.
Thore Graepel heads the Applied Games group at Microsoft Research Cambridge together with Ralf Herbrich. He obtained his PhD in machine learning from the Technical University of Berlin. His research interests include machine learning, probabilistic modeling, computational intelligence for the web, and computer games with a passion for computer go.