Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Effect of vehicle groups on heterogeneous disordered traffic flow

Akihito Nagahama, Nichika Asai, Claudio Feliciani, Xiaolu Jia, Katsuhiro Nishinari

Abstract

In heterogeneous disordered traffic, where various vehicle types operate without strict lane discipline, self-organized vehicle groups often emerge. While the formation of such groups has been recognized, their influence on macroscopic traffic dynamics remains unclear. This study investigates how the prevalence and composition of vehicle groups affect flow-density relationships in heterogeneous disordered traffic. Using trajectory data from real-world video observations, we apply three distinct Passenger Car Unit (PCU) estimation methods to construct flow-density diagrams that account for traffic heterogeneity. The analysis reveals that group proportions, i.e., the proportion of vehicles that are classified as belonging to groups, have a nonlinear and traffic-situation-dependent impact on flow characteristics. Specifically, moderate group proportions (30-60%) are associated with higher flow rates in medium- and high-density conditions, whereas proportions exceeding 50% correspond to skewed traffic distributions toward low- or high-density extremes. Comparisons between vehicle-count-based and PCU-based group proportions indicate that normalization methods significantly affect the interpretation of group dynamics, particularly when groups consist mainly of small-PCU vehicles such as motorcycles. Additionally, lower group proportions enhance flow under free-flow conditions, while the entropy-based analysis indicates that the association between entropy alone and speed is not consistently observed across traffic situations. By contrasting representative trends and extreme high-flow cases, the results further suggest that traffic under similar density and group-proportion conditions can exhibit low-efficiency and high-efficiency modes.

Effect of vehicle groups on heterogeneous disordered traffic flow

Abstract

In heterogeneous disordered traffic, where various vehicle types operate without strict lane discipline, self-organized vehicle groups often emerge. While the formation of such groups has been recognized, their influence on macroscopic traffic dynamics remains unclear. This study investigates how the prevalence and composition of vehicle groups affect flow-density relationships in heterogeneous disordered traffic. Using trajectory data from real-world video observations, we apply three distinct Passenger Car Unit (PCU) estimation methods to construct flow-density diagrams that account for traffic heterogeneity. The analysis reveals that group proportions, i.e., the proportion of vehicles that are classified as belonging to groups, have a nonlinear and traffic-situation-dependent impact on flow characteristics. Specifically, moderate group proportions (30-60%) are associated with higher flow rates in medium- and high-density conditions, whereas proportions exceeding 50% correspond to skewed traffic distributions toward low- or high-density extremes. Comparisons between vehicle-count-based and PCU-based group proportions indicate that normalization methods significantly affect the interpretation of group dynamics, particularly when groups consist mainly of small-PCU vehicles such as motorcycles. Additionally, lower group proportions enhance flow under free-flow conditions, while the entropy-based analysis indicates that the association between entropy alone and speed is not consistently observed across traffic situations. By contrasting representative trends and extreme high-flow cases, the results further suggest that traffic under similar density and group-proportion conditions can exhibit low-efficiency and high-efficiency modes.
Paper Structure (25 sections, 5 equations, 16 figures, 15 tables)

This paper contains 25 sections, 5 equations, 16 figures, 15 tables.

Table of Contents

  1. Introduction
  2. Data and Preprocessing
  3. Observed Data and Vehicle Trajectory
  4. Classification of Traffic Situations
  5. Detection of Groups
  6. Applied PCUs
  7. IP method
  8. MLR method
  9. SB method
  10. Effect of Group Proportion on the Flow-Density Structure
  11. Overview of Flow-Density Structure via Segment-Wise Medians
  12. Result
  13. Disucssion
  14. Detailed Visualization of Group-Wise Variability in Flow-Density Data
  15. Result
  16. ...and 10 more sections

Figures (16)

  • Figure 1: Traffic-observation site used for data acquisition. The downstream signal, upstream intersection, and observation section are highlighted using a black chained rectangle, black cross mark, and red dashed rectangle, respectively. The map image is based on OpenStreetMap.
  • Figure 2: Division of the observation segment into three zones (A, B, and C), arranged from downstream to upstream.
  • Figure 3: Flowchart for classifying vehicles into groups and others.
  • Figure 4: (a) Conceptual image of a group in traffic: an example where two motorcycles maintaining a leader--follower relationship are detected as a group. (b) Actual traffic scene where a group of two motorcycles is detected.
  • Figure 5: Medians of flow in PCU (IP method) for respective group proportion by number of vehicles. The percentage indicates the proportion of group vehicles in the traffic flow. $\lambda_i^{\mathrm{O}}$ is the threshold of the occurrence frequency used to detect groups.
  • ...and 11 more figures