On improving route choice through learning automata

Urban mobility is a major challenge in modern societies. Increasing the infrastructure’s physical capacity has proven to be unsustainable from a socio-economical perspective. Intelligent transportation systems (ITS) emerge in this context, aiming to make a more efficient use of existing road networks by means of new technologies. In this paper we address the route choice problem, in which drivers need to decide which route to take to reach their destinations. In this respect, we model the problem as a multiagent system where each driver is represented by a learning automaton, and learns to choose routes based on past experiences. In order to improve the learning process, we also propose a mechanism that updates the drivers’ set of routes, allowing faster routes to be learned. We show that our approach provides reasonably good solutions, and is able to mitigate congestion levels in main roads.