Traffic State Estimation Using Traffic Data from Fixed Detectors and Probe Vehicles

As availability and reliability of observed traffic data significantly affect the accuracy of traffic state estimation, traffic data from probe vehicle technique, which has a potential of network coverage, might be useful to improve the traffic state estimation. The applications of probe vehicle data are still limited as they mostly concentrated on travel time detection, such as in Chen, while some authors has used probe data to estimate O-D data, and detect incidents. As knowledge of the authors, so far, no work has applied the probe data with the macroscopic model to estimate traffic states. In fact, fundamental traffic state variables (traffic volume, space mean speed, and traffic density) have some advantages over travel time, such as, they are better to reflect traff conditions; they have a capability to convert to other variables. The objective of this study is to propose a method for integrating probe vehicle data into fixed detector data for estimating traffic states on a freeway. The Kalman filtering technique (KFT) is applied to update the state variables estimated by a macroscopic model. Firstly, the formulation of the proposed method, which considers how to treat the observation variables for the KFT in order to overcome the inconsistency of observation data, will be presented. Then, the methodology will be examined using several sets of hypothetical data under different traffic conditions