Project description

This project aims to develop:

• macroscopic traffic flow models, for signalized networks, with sufficiently high resolution to study the impact of traffic signal parameters such as splits, offsets, cycle length, phase sequence and to incorporate transit signal priority (TSP) considerations,

• performance analysis tools for signalized traffic flow dynamics with TSP constraints; optimization methods to choose signal parameters based on offline knowledge of transit schedule,

• framework for design and performance analysis of output feedback control which uses realtime transit vehicle location data and traffic flow sensor data to adjust traffic signal parameters for TSP.

Several cities are characterized by transit (trains, trams and buses) sharing signalized roadways with auto mode. With increasing interest in prioritizing transit vehicles at conflict points, it is imperative to have good models that can evaluate the tradeoffs involved in such prioritization and enable optimal operations. Macroscopic traffic flow models have enjoyed considerable success from modeling and control perspectives, starting with freeway traffic. Their application to signalized corridors has been restricted primarily to averaged models, under which the outflow from a link is equal to the effective capacity of the corresponding movement. Such approximations do not have sufficient resolution to distinguish between green and red phases, offsets, and phase sequence, all of which are relevant for arrival on green consideration in TSP. On the other hand, microscopic models are computationally too expensive for control purposes. Motivated by these, we develop high resolution macroscopic traffic flow models for signalized traffic networks and use them to design passive and active signal control strategies with TSP constraint. A salient feature of our project will be to expand the optimization based paradigm in TSP to modern developments in traffic signal control so as to obtain rigorous performance guarantees. Methodological contributions will be supplemented with Los Angeles area case study in PTV VISSIM.