Harmonization for omnidirectional progression in urban traffic control

Abstract

I describe harmonization as part of my work on distributed, knowledge-based, real-time, traffic-adaptive control of street and highway ramp traffic signals. The control system developed in prior efforts includes a two-stage learning process. The first stage optimizes the control of steadystate traffic at a single intersection and over a network of streets. The second stage of learning, not related yet to harmonization, deals with the predictivelreactive control of dynamic traffic flow in street networks. However, the control regimes at the individual intersections do not consider the time behavior of adjacent intersection control regimes. Optimal traffic flow in street networks must be with reference to signalization, phasing, and harmonization of control at all intersections. Harmonization represents the best approximation to a coordinated omnidirectional progression ("green wave"). This means that the resulting control regime produces a minimum of the sum, overall intersections, of delay times due to red lights and of unused green periods, each contributing term being weighted by the respective traffic flow values. The system has been tested in the laboratory on a range of scenarios (in terms of geometries and traffic flows) and has been found to perform as expected.