A method for using crowd-sourced trajectories to construct control-independent fundamental diagrams at signalized links

Abstract

Abstract This work presents a methodology for measuring traffic quantities from crowd-sourced trajectories for the purpose of estimating saturation flow rates. First, the observed trajectories are clustered to identify the signal programs that occur over the data collection period. Trajectories with known cycle lengths are accumulated on one space–time plot with their respective green start times at the origin, and then Edie’s generalized definitions of traffic variables are adapted to construct a full-scale fundamental diagram. The approach is tested empirically, and the averages and variances of the estimated quantities were within the expected ranges for each type of movement. Results from several movements were also verified against field measurements from a short video recording. The proposed method enables estimations of saturation flow rates over entire networks based on long-term datasets, instead of short-term field measurements. The approach is particularly advantageous for estimations under the unpredictable conditions associated with conflicting movements.