Using TraSMAPI for the assessment of multi-agent traffic management solutions

Abstract

Intelligent Traffic Management is undoubtedly a promising solution to tackle modern cities’ problems related to the growth of the urban traffic volume as it is a non-invasive approach when compared to interventions to the road network structure. Among possible solutions aiming at Intelligent Traffic Management, we believe that multi-agent systems (MASs) are the most appropriate metaphor to deal with complex domains such as road networks and traffic management and control systems. However, we feel that traffic management and control, particularly intelligent traffic control, is an issue that has not yet been addressed to its full potential. Therefore, we propose using the Traffic Simulation Management API’s multi-agent framework for multi-agent simulations over multiple microscopic simulators, as a basis for the development of intelligent policies for traffic management. We present a case study in which the advantages of cross-validation using two simulators are highlighted.