LAUSR

Abstract Assessing vehicle fuel economy in real-world driving conditions is a critical requirement to establish a reliable baseline when evaluating driver assistance systems or autonomous vehicles, where the speed profile can be optimized based on route information. Since the benchmarking is traditionally done by collecting and analyzing large amounts of data over on-road testing, virtual driver models have been developed to conduct simulation studies that allow one to understand the impact of specific driver behaviors on the vehicle speed profile. This paper presents an enhanced driver model that predicts a longitudinal vehicle speed profile based on route data, which can be calibrated with simple tests. The model extends the Intelligent Driver Model to more accurately characterize the response to stop signs, traffic lights, and other conditions typical of urban driving. The enhanced driver model can be calibrated to match the behavior of specific drivers and determine statistically-relevant distributions of model parameters.