Lane Keeping Control Using Nonlinear Model Predictive Control on Constant Speed Autonomous Car

Abstract

Lane keeping controller drives vehicle's steering to keep the vehicle driving on the track. This paper discusses lane control on a prototype autonomous car that moves at constant speed, using a nonlinear predictive control (MPC) model which is used to calculate the optimal steering angle based on lateral deviation information. The predictive lateral deviations are obtained from the linear parameter varying (LPV) model while the current lateral deviation value is obtained from the lane detection algorithm which produces a reference trajectory for the car. The lane detection uses image processing towards images captured by the camera. The real time experiment result shows that the proposed controller could keep the prototype to stay on track until the curvature of 0.27 m-1 with the maximum lateral deviation of 8.86 cm.