The Neural Network Direct Inverse Control of Four-wheel Steering System

The motorcycle model which has two degrees of freedom (2-DOF for short) is adopted to study the characteristics of four-wheel steering (4WS for short) vehicle, and the linear 2-DOF dynamic equation of 4WS vehicle is constructed in this paper. A neural network direct inverse controller is designed to control the steering system of 4WS through the offline identification and online learning processes. In contrast to other control methods, the neural network direct inverse controller is more effective in controlling the steering angle of rear wheel to actualize the slip angle minimization and to improve both the mobility at low speed and the handling stability at high speed.