Racing Driver Modeling Based on Driving Behavior

Abstract

The driver model is an important link in the research of shared autonomy control. In order to simulate the driver’s handling characteristics in the complex human-vehicle-road closed-loop system, the driver model is required to accomplish the driving operation under specific working conditions. In this paper, a lateral-longitudinal combined racing driver model is designed. The lateral control model adopts the preview model with far and near viewpoints and the dynamic velocity controller is added into the longitudinal control model to obtain the expected speed of the target trajectory. Finally, the racing driver model proposed in this paper is validated through simulation on track conditions of FSAE. In the given conditions, the result shows the racing driver model outperforms the typical driver model in lateral path tracking and the speed of racing driver model is higher than typical model on straight and corners. Meanwhile, the representation of driving skills is a key step to enhance the adaptive control of vehicles in the future. The control parameters can be adjusted according to the driver’s skill information to make the vehicle control system adapt to the driver’s skill level. This paper introduces the method of driving skill recognition based on wavelet transform and Lipschitz singularity detection theory and the preliminary test results prove the feasibility of using this method to characterize the driver’s operating skill level.