Estimating Queue Length under Connected Vehicle Technology

Abstract

With the emergence of connected vehicle technology, the use of probe trajectory data to estimate queue length has recently received considerable attention. Unlike data collected by loop detectors, probe trajectory data can provide a lower bound on the queue length even if the market penetration rate is low. An event-based method is developed: it uses both probe trajectory and signal timing data to estimate queue length, and the estimation accuracy under different market penetration rates is examined. A data fusion method is developed: it combines probe trajectory data and loop detector data, and the situation in which the data fusion method is expected to work well is investigated. Case studies are conducted with microscopic simulation data, and some observations are made.