Estimating real-time traffic state of holding vehicles at signalized intersections using partial connected vehicle trajectory data

Abstract

Emerging connected vehicle (CV) technologies offer unprecedented opportunities to estimate various traffic states, enhancing traffic management and control. Among these states, a particularly critical yet underexplored one is the number of holding vehicles—vehicles that, based on their projected trajectories using cruise speed, should have been discharged at any instant of interest but are instead impeded and remain undischarged. Accurately estimating this quantity is essential for real-time traffic state monitoring and control, as it directly reflects the effectiveness of traffic flow at intersections. However, the prolonged transition period implies a mix of CVs and non-connected vehicles (NCs) within transportation networks, resulting in incomplete traffic information. To address this challenge, this paper proposes a generic and fully analytical CV-based holding vehicle (CVHV) model to estimate the number of holding vehicles at any instant of interest, relying solely on partial CV trajectory data. The CVHV model accommodates any signal plans, CV penetration rates, and traffic demands. Two sub-models, CVHV-I and CVHV-II, are derived to account for different holding vehicle patterns at any instant of interest falling within the effective red or green of a signal group, respectively. Each sub-model handles various holding vehicle patterns, including holding vehicle components such as stopped holding CVs and NCs, and moving holding CVs and NCs. Comprehensive numerical experiments in VISSIM validate the effectiveness of the CVHV model under varying volume-to-capacity ratios, CV penetration rates, and signal timing configurations. Its practical applicability is further demonstrated using the real-world Next Generation Simulation dataset. Additionally, the application of the proposed model to estimating the real-time total number of vehicles in a lane and to a simple illustrative example of CV-based adaptive signal control highlights the significance of accurately estimating the traffic state of holding vehicles.