A methodology for calibrating road link travel time functions using data from driving simulator experiments

In road network traffic operations and management, travel time is a widely used performance indicator playing a significant role in transportation systems modeling. Link cost functions explain the behavior of traffic variables (travel time in particular) over the full range of road traffic conditions (in particular flow rate). Travel time estimation is a very interesting topic in traffic analysis activity. Estimating travel time using field data from existing devices or measurement procedures (such as inductive loop sensors, license plate recognition systems, etc.) is attractive. However relatively few works have been done on practical aspects of estimating link performance functions from field data; in fact, usually paired travel time-flow data is both difficult and expensive to obtain. This paper considers the estimation of link travel time functions from data collected by experiments in a driving simulator and proposes a comparison with the link travel time functions estimated using loop detector data. The study is motivated because many research activities have demonstrated that driving simulators are good tools to improve the driving behavior knowledge, suitable alternatives to field studies for instance to analyze drivers' behavior with reference to situations where detection and control of the descriptive variables (measurements under controlled traffic conditions) are very difficult to make. The analysis context was a road segment belonging to a rural two-way two-lane road. The experiments have been conducted at the Transportation Laboratory of the Department of Structural and Transportation Engineering (University of Padova) using STSoftware\uae five-screen configuration driving simulator. The results obtained appear interesting and confirm as driving simulator can be used in order to develop deeper analysis of road network performance