1. Traffic arrival pattern estimation at urban intersection using license plate recognition data.
- Author
-
Li, Min, Tang, Jinjun, Chen, Qun, and Liu, You
- Subjects
- *
TRAFFIC patterns , *AUTOMOBILE license plates , *TRAVEL time (Traffic engineering) , *TRAFFIC signs & signals , *TRAFFIC engineering , *INTELLIGENT transportation systems - Abstract
Traffic arrival pattern is an indispensable part of the traffic state analysis and optimization of signal control schemes at the urban intersection. In some previous research relevant to the traffic parameters at the intersection, such as the queue length estimation, delay estimation, and vehicle trajectory reconstruction, the vehicle arrival patterns are assumed to obey a Normal or Poisson distribution. However, the actual traffic arrival patterns at the intersection are dynamic and more complex. License plate recognition (LPR) data containing rich vehicle trajectory information are promising in data-driven applications of traffic control and management. We propose a framework based on the LPR data, to infer the dynamic traffic arrival pattern at the intersection. The proposed integrative data-driven framework consists of three main parts, a method for inferring traffic signal cycle parameters, deriving the cycle start time and cycle length from time headway sequences; a link travel time estimation method, extracting the valid travel time at the selected time interval, and an arrival pattern estimation model based on the above modules to perform the second-level estimation at the downstream intersection. The unique characteristics of LPR data are exploited and utilized in this study. To validate the proposed method, the LPR data from two adjacent intersections in the city of Changsha in China are used. Numerical results reveal that the proposed framework can achieve satisfactory estimation under different traffic scenarios. The findings in this study can be extended for supporting efficient traffic control applications. • A statistical framework to infer the fine-grained traffic arrival pattern is proposed. • The signal cycle parameters are inferred by exploiting the features of the time headway sequences. • The valid travel time is extracted at the selected time interval based on the Gaussian mixed model. • The proposed framework could achieve satisfactory performance in different traffic scenarios. [ABSTRACT FROM AUTHOR]
- Published
- 2023
- Full Text
- View/download PDF