Optimization-Based Queue Estimation on an Arterial Traffic Link with Measurement Uncertainties

Advanced monitoring and control of arterial road traffic network operations requires accurate knowledge of current and predicted performance measures on the network. Recently studied signal control algorithms, for example, use the lengths of vehicle queues for each turning movement to determine how subsequent signal cycles should be distributed into phases. This article presents a queue estimation procedure that can integrate measurements from classical count or occupancy sensors into a single physical model of general link state and queue length in particular. The authors show how realistic data from an arterial link can be used to estimate the current or recent state of this link by manipulating the initial and boundary conditions used in an explicit solution to the Moskowitz (cumulative number of vehicles) formulation of the Lighthill-Whitham-Richards (LWR) partial differential equation. The authors demonstrate the results of this estimation procedure using various sensor configurations extracted from data and ground-truth vehicle trajectories taken from the NGSIM community’s Lankershim Blvd data set.