Evaluating Driving Styles by Normalizing Driving Behavior Based on Personalized Driver Modeling

Abstract

Driving style evaluation is important for vehicle calibrations and intelligent transportation. In this paper, we propose to quantitatively evaluate driving styles by normalizing driving behavior based on personalized driver modeling. First, a personalized driver model is established for each driver to be evaluated by using the neural network, e.g., the radial basis function, and real-world vehicle test data, with respect to vehicle and road situations. Second, the established driver model is employed to perform the simulated standard driving cycle test for driving behavior normalization, where the desired speed profile is adopted from the standard driving cycle test, e.g., federal test procedure-75. Third, based on the energy spectral density analysis on normalized behavior, an aggressiveness index is proposed to quantitatively evaluate driving styles. Finally, this index is applied to detect abnormal driving behavior. Simulations are conducted to verify the effectiveness of the proposed scheme.