Study of Traffic Flow Controlled with Independent Agent-Based Traffic Signals

Abstract. Dealing with urban traffic is a highly complex task since it involves the coordination of many actors. Traditional approaches attempt to optimize traffic signal control for a particular vehicle density; the main disadvantage lies in the fact that traffic changes constantly. Managing traffic congestion seems to be a problem of adaptation rather than of optimization. In this work we present an agent-based traffic simulator which represents a traffic grid with two-way roads of three exclusive lanes per direction, with intersections regulated by signals. We study the repercussions on traffic flow of simple parametric behaviours when each light operates independently. A dominance analysis is applied to compare the strategies.