Dynamic traffic assignment model considering drivers' unfamiliarity with network layout

Abstract

Most existing dynamic traffic assignment models assume that drivers have sufficient knowledge on roadway networks. However, past experiments have shown that drivers' familiarity with the network layout is an essential component in route selections. In this paper, the concept of recognition level is defined to categorize drivers based on their unfamiliarity of the network and of the alternative routes between origins and destinations. Each catalog is assigned a specific utility function that is dependent on travel time, length of route and recognition parameters. Drivers' route choice behavior is determined by these specific utility functions. A sample network is first employed to test the feasibility of the proposed model, and the result complies with the specified travel patterns. After that, a real network near downtown Houston is used to further test the proposed model. An experiment is conducted based on the information collected from an on-site survey and the on-line realtime traffic map from Houston TranStar. In order to validate the necessity of the proposed model, a control experiment is carried out with all parameters being set in the same way as the designed experiment except that drivers are assumed to be fully familiar with the network layout and alternative routes. Test results show that the proposed model better fit the real case.