Adaptive Identification Method of Vehicle Modeling According to the Fluctuation of Road and Running Situation in Autonomous Driving

Abstract

In recent years, study on autonomous driving has been actively conducted. The model-based control is used as one of control methods of autonomous driving. The model-based control in automated driving requires simple and accurate vehicle model that can accurately represent the state of the vehicle and has low processing load on computer memory. Therefore, we represented the modeling error due to fluctuation of road and dynamic characteristics of the vehicle as the deviation of the front wheel steering angle of the kinematics bicycle model which is a simple model. We propose a method to adaptively identify the vehicle model while estimating the modeling error in real time using Neural Network. In this paper, we apply the vehicle model including modeling error to the model predictive control as an example of model-based control, and verify the usefulness of this study in simulation using CarSim.