Understanding and Modeling the Human Driver Behavior Based on MPC

Abstract

Better understanding driver behavior is essential for autonomous vehicles, as well as simulation, evaluation and optimization of a driver-vehicle closed-loop system. In this paper, the human driver behavior is first discussed in the form of perception, decision, and execution. In the following, the modelling of the driver behavior is proposed based on the preview-follower theory. The driver model includes three parts which are perception module, decision making module and execution module. The decision making module contains longitudinal control using proportional-integral-derivative (PID) controller and lateral control using model predictive control (MPC) method. Moreover, the response of muscle and nerve is modelled in the form of execution module. Typical simulations are carried out using vehicle dynamics software veDYNA and indicate that the proposed driver model is suitable for representative path following at varying vehicle speed.