Model Predictive Control for Reliable Path Following with Application to the Autonomous Vehicle and Considering Different Vehicle Models

Abstract

In this paper, a reliable path following approach based on Model Predictive Control (MPC) and using different vehicle models are developed to: 1) minimize the lateral distance between the vehicle and a reference path, 2) minimize the heading error of the vehicle, and 3) limit the rate of steering inputs to their saturation values and produce smooth motions. The proposed MPC is implemented in MATLAB/Simulink using CasADi block in which the associated Optimal Control Problem (OCP) is discretized using the direct multiple-shooting method. Results show that designed controllers pulled the system back to the predefined reference path, and tracking performances were satisfactory. The effect of changing the prediction horizon and road friction coefficient at low/high speeds is discussed.