A Real-Time Model Predictive Control Approach for Autonomous Active Steering

Abstract

The problem of controlling the front steering to stabilize a vehicle along a desired path is tackled in this paper. Although a Non-LinearModel Predictive Control (NLMPC) approach can achieve good performance and constraints fulfillment, its computational burden does not allow a real-time implementation. In order to decrease the complexity of the controller, in this paper we propose a suboptimal MPC scheme based on successive on-line linearizations of the non-linear vehicle model. The method stems from an accurate analysis of the vehicle nonlinearities, the constraints and the performance index in the optimal control problem. The simulation results show a significant reduction of the controller complexity, with a small loss of performances respect to a NLMPC controller. The suboptimal MPC control policy is compared to the control policy of a robot driver from Ford Motor Company.We show that better performance can be achieved with a smaller control effort, without violating vehicle physical constraints, by using a systematic control design procedure.