Using traffic data of various types in the estimation of dynamic O-D matrices

Dynamic Traffic Assignment (DTA) has become a main component of modern traffic control centres. To calibrate a DTA model the observations from the field are required. There has been increasing number of sensors and technologies which can provide these data. In this paper we briefly describe these sensors and elaborate on the various traffic data types that are used in dynamic demand calibration. The methodology for dynamic demand calibration using various data types is presented. We test several of these data types and provide the comparison of their effectiveness. As this is an optimization problem we test three derivative-free optimization algorithms and provide their comparison. The advantages and disadvantages of different traffic data as well as of optimization algorithms are discussed.