Inference of signal phase and timing with low penetration rate vehicle trajectories

Traffic signals are a crucial component of urban traffic networks, and signal phase and timing (SPaT) information serves as an essential input for various urban traffic operational applications. Obtaining SPaT information on a large scale is challenging due to the diversity of traffic signal controllers from different manufacturers and jurisdictions. With the advent of broadly defined connected vehicles, vehicle trajectories can be leveraged to estimate SPaT information since they are directly controlled by traffic signals. Although some existing studies have proposed methods for estimating SPaT information using vehicle trajectory data, most are limited to fixed-time traffic signals. To address this limitation, this paper proposes a suite of SPaT inference algorithms applicable to both fixed-time and responsive signals. With only low penetration rate vehicle trajectory data as input, the inference program can estimate the complete SPaT information for traffic signals with fixed cycle lengths and the average cycle/splits for those with time-varying cycle lengths. The proposed method is validated through case studies at real-world intersections. • Develop a signal phase and timing inference method with low penetration rate vehicle trajectories as the only input. • Ensure applicability across various traffic signal controllers, including both fixed-time and responsive. • Introduce efficient algorithms for estimating key signal parameters such as time-of-day plans, cycle lengths, and green splits. • Validate the proposed method using real-world data and offline traffic signal reports.