Estimating traffic volumes for signalized intersections using connected vehicle data

Zheng, Jianfeng; Liu, Henry X. · 2017 · OpenAlex-citations

DOI: 10.1016/j.trc.2017.03.007

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 challenge of estimating traffic volumes at signalized intersections using connected vehicle (CV) data under low market penetration rates. While CV technology offers the potential to replace conventional vehicle detectors for signal operation, existing research largely assumes high penetration rates (e.g., 25%) or relies on simulated data. The authors aim to fill this gap by developing a method to estimate traffic volume—a critical input for signal optimization—using sparse GPS trajectory data from CVs or navigation devices. This approach is motivated by the need to support traffic management agencies in evaluating and operating signals efficiently as CV deployments scale, paving the way for detector-free signal control. The methodology models vehicle arrivals at intersections as a time-dependent Poisson process, which accounts for signal coordination effects. The estimation problem is formulated as a maximum likelihood problem based on observed CV trajectories. The authors distinguish between two types of trajectory data: vehicles that stop at the intersection, which provide direct information on queue lengths and arrival counts, and vehicles that pass through without stopping, which provide upper bounds on arrivals. To solve for the mean arrival rate, an Expectation Maximization (EM) algorithm is derived to handle the partial information provided by non-stopped vehicles. The model incorporates signal phase and timing data to map trajectories to signal clock time, allowing for the calculation of time-dependent arrival factors. The proposed algorithm was validated through two case studies using real-world data. The first study utilized CV data from the Safety Pilot Model Deployment (SPMD) project in Ann Arbor, Michigan, involving approximately 2,800 equipped vehicles. The second study used vehicle trajectory data from a commercial navigation service in China. Performance was evaluated against benchmark data manually collected and data from loop detectors. The results demonstrated that the mean absolute percentage error (MAPE) of the volume estimates ranged from 9% to 12%. The study confirmed that the method effectively leverages limited trajectory data to infer overall traffic volumes, even when penetration rates are as low as 3–12%. The significance of this work lies in its demonstration that accurate traffic volume estimation is feasible with low-penetration CV data, challenging the assumption that high penetration rates are necessary for CV-based signal control. By providing a reliable method to estimate key inputs for signal optimization, the approach offers a practical tool for traffic agencies to fine-tune signals periodically. This research serves as a foundational step toward fully adaptive, detector-free signal control systems, reducing reliance on costly fixed-location detectors and improving the efficiency of urban traffic networks.

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