Agent-Based Traffic Control: a Fuzzy Q-Learning Approach

Network modelling and simulation is an important approach to implement effective traffic control strategies and management policies for urban mobility. With much advance in communication and information technologies, and as ubiquitous computing becomes part of most people´s lives, both infrastructure and user experiment a new dimension of their relationship that demands for a proper behaviour from control and management systems. In this paper we use the autonomous agent metaphor to implement intelligent control through a fuzzy Q-learning approach. The controller decision-making mechanisms and some preliminary simulation experiments are set up so as to assess the feasibility of our model to cope with urban traffic coordination and management.