An Explanatory Approach to Driver Modeling
Dr. Daiheng Ni, Principal Investigator
Assistant Professor, Department of Civil and Environmental Engineering
University of Massachusetts/Amherst
To reduce the high cost and repeated investment in calibration, new principles are proposed for the development of a new generation of transportation modeling and simulation tools. A fundamental explanatory principle is postulated in this research. Unlike the conventional descriptive principle which tries to describe the subject system by fitting observed data, the explanatory principle captures the essential mechanisms that drive the behavior of the subject system. As a first step, this project will develop a rational driver model. The driver will be modeled as an autonomous intelligent agent that is motivated by goals which drive the agent’s behavior. Inputs to the driver model include information from the driving environment and vehicle feedback. Outputs of the driver model include the desired level of acceleration/deceleration and steering. The behavior exhibited by the driver is modeled as decisions at three levels: the global-level decision concerns navigation (i.e. choice of route to reach the destination, anticipated travel times, and level of desired safety), the local-level decision concerns operation (i.e. selection of lanes, interaction with neighboring vehicles, and response to road geometry and traffic control devices), and the vehicle-level decision concerns control (i.e. the desired acceleration, deceleration, and steering). This research addresses the UMass Transportation Center theme “Improving Transportation Mobility and Safety with Innovative Technologies and Strategies” and this research responds to a national priority by contributing to traffic congestion and highway safety. Results of this research will be particularly useful for developing powerful tools to study performance characteristics of drivers, as well as to develop innovative safety and mobility solutions for them.