Traffic demand estimation for lane groups at signal‐controlled intersections using travel times from video‐imaging detectors

Abstract

The purpose of this study is to present a new method for lane‐based traffic demand estimation using travel times from video‐imaging detectors. The method overcomes the following two shortcomings of loop‐detector‐based algorithms: the fact that the actual demand is unknown when detectors are located upstream from the stop lines within a short distance; and the difficulty in calculating the ratio between streams in different lane groups if detectors are located at the upper reaches of the links. First, the authors analyse a variety of travel time patterns and introduce the concept of a virtual cycle that satisfies the criteria that all vehicles entering into a link in one virtual cycle have just departed from a downstream stop line within a single signal cycle. Next, the authors improve the travel time reduction rate model for queued vehicles in each cycle, and enhance the algorithms to estimate the lane‐based traffic demand under different conditions. Finally, all parameters are calibrated and the new models are evaluated. The results show that: the maximum, minimum and average deviations over 12 cycles are 38.50, 0.02 and 16.19%, respectively. The findings in this study have potential applicability for use in traffic control systems, especially where oversaturated conditions are present.