Sensing technologies for traffic flow characterization: From heterogeneous traffic perspective

Ayaz, Shehzad; Khattak, Khurram; Khan, Zawar; Minallah, Nasru; Khan, Mushtaq; Khan, Akhtar · 2022 · Crossref

DOI: 10.5937/jaes0-32627

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the critical need for accurate traffic flow characterization to support Intelligent Transportation Systems (ITS), particularly in environments with heterogeneous traffic. The authors argue that existing sensing technologies—categorized as manual counting, intrusive sensors, and non-intrusive sensors—possess significant limitations when deployed in congested or mixed-traffic conditions. Intrusive sensors, such as pneumatic tubes and inductive loops, are labor-intensive, expensive to maintain, and often fail to provide reliable data when vehicles stop or move slowly. Non-intrusive sensors, including radar and acoustic devices, are easier to install but suffer from sensitivity to meteorological conditions and roadway geometrics, often failing to distinguish between diverse vehicle types like pedestrians, two-wheelers, and animal-driven carts. To overcome these deficiencies, the study proposes a computer vision-based solution capable of characterizing traffic flow across all conditions, including congested and heterogeneous scenarios. The methodology involves a comparative analysis of existing sensing technologies followed by the development and field testing of a computer vision system. The proposed solution was evaluated on a complex road configuration consisting of a two-way, multi-lane road with three U-turns. Unlike traditional sensors that require multiple units for complex layouts, this system utilized a single camera to capture video data. The computer vision algorithms processed this data to extract detailed traffic flow parameters, including vehicle count, speed, spatial and temporal densities, trajectories, time/distance headways, and heat maps. The system was specifically designed to detect a wide range of road users, including pedestrians, two- and three-wheelers, and human or animal-driven carts, which are often missed by conventional sensors. The results demonstrate that the proposed computer vision solution successfully measured the aforementioned traffic flow parameters with 87% accuracy. The system proved capable of handling the complexities of heterogeneous traffic, providing detailed data on diverse vehicle types and movement patterns that intrusive and non-intrusive sensors could not reliably capture. The study highlights that while intrusive sensors offer high accuracy in homogeneous conditions, their performance degrades significantly in congestion. Similarly, non-intrusive sensors, though easier to deploy, lack the granularity required for mixed-traffic environments. The computer vision approach, by contrast, provided comprehensive characterization without the installation disruptions or environmental sensitivities associated with hardware-based sensors. The significance of this work lies in its contribution to the advancement of ITS methodologies for regions with heterogeneous traffic patterns. By demonstrating that a single-camera computer vision system can accurately characterize complex traffic flows, the paper suggests a scalable and cost-effective alternative to traditional sensor networks. This approach enables more precise validation and calibration of traffic simulation software, such as VISSIM and Paramics, leading to better urban planning and traffic management. The findings imply that computer vision is emerging as the optimal solution for traffic monitoring, offering superior adaptability and data richness compared to previous generations of sensing technologies.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• traffic density