|Institution:||Delft University of Technology|
|Keywords:||Traffic state estimation; Floating car data; Extended Kalman Filter|
|Full text PDF:||http://resolver.tudelft.nl/uuid:8709945c-6102-4cc9-a791-8cc42464c7e7|
Traffic state estimation is an important element in traffic management systems. In this research a freeway traffic state estimation methodology is proposed which allows to incorporate the information and uncertainties of heterogeneous data-types, namely loop-detector data and floating car data. The loop-detector data provides estimation of speed, flow and indirectly density, while the floating car data only provides a speed estimate. An Extended Kalman Filter (EKF) is used to combine the observations (data) with a traffic flow model. This traffic flow model, the information in the floating car data is able to affect the estimation for speed, flow and density. The EKF is able to incorporate the uncertainties in the traffic flow model and data-based estimations. This is especially important, as the estimations based on floating car data are shown to be dependent on the traffic conditions and the fraction of vehicles which are observed (penetration rate). Therefore, in the proposed methodology the uncertainties assigned to the floating car data-based estimation are dependent on the estimated traffic conditions and penetration rate.