1373452 - Intersection Queue Length Estimation using High-Resolution Event Data

Queue length is one of the most important metrics required for the performance assessment of signalized intersections because of its simple interpretability and correlation with other important signal performance measures. However, the current methodology of estimating queue length in the literature suffers from several drawbacks, including unstable estimation and the requirement of multiple data sources. Moreover, for single-channel advance detection, which is a common detection scheme for signal control in many US cities, manual calibration of threshold parameters is required. Owing to these drawbacks, previous queue length estimation methods based on high-resolution event data from single-channel advance detectors are not readily transferable to other intersections, time periods, or detector configurations. To bridge these gaps, this study proposes a cycle-based maximum queue length estimation method based on two steps: a) the empirical observations of breakpoints in the time gap between successive actuations; and b) the identification of queue status for all detector actuations in a cycle. Maximum queue length for cycles with long queues is estimated based on the saturation flow rate and the trajectory of the last vehicle in the queue. The proposed methodology is implemented using high-resolution event data collected from two study intersections along one of the busiest corridors in Tucson, Arizona. For validation, the actual maximum queue length is determined from ground-truth video observations. Results show that the proposed empirical approach achieved high accuracy in identifying both the breakpoints as well as the cycles where queue length extends beyond the advance detector. For the two study intersections, queue lengths were estimated with mean absolute percentage errors of 14.77% and 15.1% and mean absolute errors of 25 ft and 42.5 ft, respectively. The results showed significant improvements in queue length estimation from single-source single-channel detection data when compared to similar methods in the literature. It is anticipated that the proposed method can support transportation agencies accurately estimate queue length at intersections with single-channel advance detection without the need for manual field data collection and without installing lane-by-lane detection.