A Multiagent System for Helping Urban Traffic Management

Abstract

This paper describes the MASHGREEN DM prototype following three goals: 1) The identification and implementation of the tasks related to urban traffic control that use deep reasoning mechanisms as main tools for their resolution. The chosen model for explaining urban traffic behaviour is a qualitative model, which includes among its main features their low temporal and spatial computational costs. 2) The definition of a functional architecture with soft real time constraints that integrates the developed tasks. A specilized component named agent, composed by four functional modules executes every task. These modules are the following ones: communication protocols, methods (agent specilization), data space and control. This architecture verifies that the execution performances of every agent are not compromised by the inclusion of new agents in the system, or by the interactions due to the active system agents. 3) The implementation of a prototype in a high performance computational architecture, a Beowulf computer system.