ÖGAI Journal 25/3 1 Artificial Intelligence Techniques in Traffic Control Stefan Rass, Kyandoghere Kyamakya Department of Informatics-Systems Transportation Informatics Group Alpen-Adria-Universität Klagenfurt stefan.rass@uni-klu.ac.at kyandoghere.kyamakya@uni-klu.ac.at Abstract Traffic Control is an area of increasing importance. During the last few decades, traffic demand has increased so rapidly that classical control mechanisms often exhibit serious limitations, thus calling for a new paradigm and decentralized solu- tions. Artificial intelligence techniques have turned out to be a highly promising field for solving the problems in this area, including large-scale decentralized traf- fic management done by agents. We present an overview of some recent ap- proaches in traffic control with a particular focus on the use of AI technology. The described technologies and systems do either act (almost) on their own (with hu- man support only in case of emergency) or may support human traffic managers. We also review some case studies, showing that the resulting systems have be- come mature enough to be applied to real-world systems and phenomena. Introduction The term traffic control summarizes a large variety of problems, all aiming at resolving dissatisfaction of individuals within a transportation system. When we are required to travel from A to B, we normally try to choose our method and route of travelling according to minimization with respect to two variables: travel time and travel cost. However, these requirements are very often somewhat incompatible, since fast travelling is usually tied to high costs, while cheap transportation is often slow. In many cases, particularly in car traffic, the driving route is chosen to be the shortest path between the origin and the des- tination. Although an individual will most unlikely select the truly shortest path upon non- technical means, the emerging use of navigation systems renders this assumption more and more realistic. The resulting problem is the simultaneous choice of the same route- sections by very many travellers and this in turn implies a high probability of traffic jam. Particular route (road) sections may be decided to be optimal either because of length or possible or allowed travel speed on that section. In the latter case, the section is as- sumed to offer fast transportation, hence is chosen by very many individuals. The goal of traffic control techniques is a systematic influence of the transportation network, such that vehicle streams are routed in order to achieve optimal link utilization. That is, a good traf-