Urban Traffic Control system using self-organization

Abstract

This paper presents a Urban Traffic Control (UTC) system, inspired to techniques exploited by social insects to coordinate themselves and specialize their behaviour, without any centralized coordination or explicit communication. Local traffic is handled at an intersection by a controller, executing simple local reactive rule-based policies. Every intersection controller chooses on its own which policy to use, according to stimuli perceived from the environment through sensors, thus being influenced by other controllers indirectly. The execution of simple reactive local policies lets an overall traffic control to emerge, with complex behaviours not defined a priori and unaware to the single controller. Simulations demonstrate that the exploiting of emergent behaviours is desirable, as the system outperforms traditional traffic control methods. Dynamic specialization makes the system able to cope with traffic variation and to perform traffic shaping in a positive way.