Using Asymmetric Theory to Identify Heterogeneous Drivers’ Behavior Characteristics Through Traffic Oscillation

Qian Wan,Guoqing Peng,Felipe Inomata,Qianqian Liu,Yu Zheng,Zhibin Li

doi:10.1109/access.2019.2930762

Abstract

This paper applies the asymmetric driving theory to capture driving characteristics of car-following behavior throughout traffic oscillation. Unmanned aerial vehicle (UAV) was employed to record videos near a bottleneck on an expressway in Nanjing, China, from which authors used advanced image processing technology to track and extract the high-fidelity vehicles trajectory data for a microscopic analysis in this study. First, with analyzing the individual vehicle trajectory throughout oscillation, authors find that the driving characteristics of drivers are heterogeneous but show several consistent features: before and after experiencing oscillation, relatively more drivers tend to maintain aggressive driving, i.e., 49.7% are originally aggressive (OA) and 37.2% are later aggressive (LA). Second, the statistical analysis indicates some representative characteristics of Chinese drivers: the polarization (aggressive or timid) is more significant before oscillation (OA 49.7%, originally Newell ON 18.9% and originally timid OT 31.4%) but changing to relative equilibrium after oscillation (LA 37.2%, later Newell LN 33.8% and later timid LT 29%). Finally, the authors also find that the type and intensity of each driver's reaction in oscillation are related to the characteristics he or she have before encountering oscillation. These findings of asymmetric driving behavior evolution in this paper may help to explain the causes of hysteresis and unstable traffic flow phenomena.

Highlights

In congested traffic condition, drivers are largely affected by movements of nearby vehicles
Many scholars have noticed that the growth, propagation and dissipation of traffic oscillation are closely related to driver behavior characteristics
OF ANALYSIS The trajectory screening criteria for the pair samples to study car-following behavior analysis as follow: any pair of vehicles which have complete trajectories within the time and space of this study, i.e., they travel on the same lane all the way when through the measured segment without any lane changer insertion

Summary

Introduction

Drivers are largely affected by movements of nearby vehicles. Many scholars have noticed that the growth, propagation and dissipation of traffic oscillation are closely related to driver behavior characteristics. Theoretical studies [1], [2] analyzed the linearity of car-following models and attributed the propagation of oscillation to the over-reaction of car-following behaviors. Gipps developed a car-following model based directly on driver behavior and expectancy as regards downstream traffic [3]. Many researches on driving behavior are based on simplified car-following theory that proposed by Newell [4], which considers reaction time and congested spacing. Numerous driver behavior models fail to adequately describe the complex driving behaviors in congested traffic, as they are pervasively based on the assumption of symmetry during acceleration and deceleration processes. Asymmetric driving behavior is related to hysteresis phenomenon which indicates the retardation in speed

Objectives

Results

Conclusion