Design and experimental validation of linear and nonlinear vehicle steering control strategies

Abstract

This paper proposes the design of three control laws dedicated to vehicle steering control, two based on robust linear control strategies and one based on nonlinear control strategies, and presents a comparison between them. The two robust linear control laws (indirect and direct methods) are built around M linear bicycle models, each of these control laws is composed of two M proportional integral derivative (PID) controllers: one M PID controller to control the lateral deviation and the other M PID controller to control the vehicle yaw angle. The indirect control law method is designed using an oscillation method and a nonlinear optimisation subject to H _∞ constraint. The direct control law method is designed using a linear matrix inequality optimisation in order to achieve H _∞ performances. The nonlinear control method used for the correction of the lateral deviation is based on a continuous first-order sliding-mode controller. The different methods are designed using a linear bicycle vehicle model with variant parameters, but the aim is to simulate the nonlinear vehicle behaviour under high dynamic demands with a four-wheel vehicle model. These steering vehicle controls are validated experimentally using the data acquired using a laboratory vehicle, Peugeot 307, developed by National Institute for Transport and Safety Research – Department of Accident Mechanism Analysis Laboratory's (INRETS-MA) and their performance results are compared. Moreover, an unknown input sliding-mode observer is introduced to estimate the road bank angle.

Keywords:

• vehicle dynamics
• experimental validation
• steering control
• switching control
• LMI optimisation
• sliding-mode control
• road bank angle estimation

Notes

1, indirect method: two-degree-of-freedom PID multi-controllers; and 2, direct method: robust switching two-PID controllers via LMI–LQR.