EXPERIMENTAL INVESTIGATION OF THE DRIVER'S BEHAVIOR AT UNCONTROLLED INTERSECTIONS IN CROSS TRAFFIC FLOWS

Abstract

Problem. When modelling traffic flows through intersections of the network of streets, the most difficult task is to model driver's behavior, since it defies precise mathematical description, it depends on almost unlimited number of factors and can significantly change even for the same driver over a fairly short period of time, for example, one trip. Goal. The goal is to obtain the values of Cd and to build the histogram of distribution of these values probabilities for the chosen type of maneuver. Methodology. The method of modeling driver's behavior at uncontrolled intersections in cross traffic flows has been investigated in the paper. The analysis and processing of video recording of cross traffic flows at the four-way intersection chosen on the network of streets of the city of Kharkiv has been performed. An improved method of «boundary intervals» as a «sliding boundary interval» has been proposed for modeling the driver's behavior. Results. As a result of processing the video recording of the flow motion at uncontrolled intersections, the histogram of the distribution of probabilities of the determination coefficient values at cross traffic flows has been built [11, 12]. With the help of such histograms, the Cd values for each car in the flow can be specified using any known random number generator. Originality. The method stated will take into account both the individual characteristics of drivers and the impact of objective factors on their behavior; the proposed method of recording drivers when modeling the movement of traffic flows will increase the accuracy of modeling, which in turn will allow us to specify a number of results that we obtain by modeling, for example, the calculation of the traffic lane capacity. Practical value. The proposed method of simulating the driver's behavior in the models of traffic flows makes it possible to approximate the accuracy of simulation to the real variety of behavior of drivers, and, consequently, to increase the adequacy of models.