University of California Transportation Center

In order to develop efficient control strategies to improve traffic conditions on freeways, it is necessary to know the state of the freeway at any point in time and space. Using data collected from stationary detectors –such as loop detector stations– the density field can be currently reconstructed to a certain accuracy. Unfortunately, deploying this type of infrastructure is expensive, and its reliability varies. This article proposes and investigates new algorithms that make use of data provided by mobile sensors, in addition to that collected by stationary detectors, to reconstruct traffic flow. Two approaches are proposed and evaluated with traffic data. The first approach is based on data assimilation methods (so-called nudging method) and the second is based on Kalman filtering. These approaches are evaluated using traffic data. Results show that the proposed algorithms appropriately incorporate the new data, improving significantly the accuracy of the estimates that consider loop detector data only.