Investigation of car-following models in disordered traffic using trajectory data obtained from unmanned aerial vehicles

Abstract

ABSTRACT Car-following models are the cornerstones of microscopic traffic simulation tools and intelligent transportation systems, but the applicability of car-following models to disordered traffic have not been investigated in detail with longer trajectory dataset. To address this gap, two car-following models namely, Intelligent Driver Model (IDM) and Full Velocity Difference Model (FVDM) are calibrated using trajectory data collected on an urban arterial road in Chennai, India using Unmanned Aerial Vehicles. The raw data are smoothed for noise removal and the car-following pairs are identified based on the lateral overlap and following duration. The models are calibrated by minimizing the deviations between the observed and simulated longitudinal gaps between leader and follower using genetic algorithm. The obtained errors are between 2.5% and 19.5%, which are comparable with standard ranges of error reported in literature. The optimal parameter values represent the distinct characteristics of disordered traffic in comparison with the homogeneous traffic.