Model Predictive Control Approaches for Lane Keeping of Vehicle

摘要

The objective of this paper is to describe workflow and analysis of the implementation of two different approaches of Model Predictive Control as applied to lane keeping feature for a passenger vehicle in virtual environment. It compares the performance of the approaches in the model-in-loop scenario which could serve as a first step towards the hardware-in-loop validation of MPC implementation and its realization on target embedded hardware thereafter that goes onboard. It covers the development of simplified model for electrical power steering system followed by modeling the lateral dynamics of a vehicle as a Linear Parameter Variant state space model in time domain for a commonly followed speed range. This is followed by synthesis of controller using the two MPC approaches namely, (i) Quadratic Programming approach, (ii) Explicit MPC using multi-parametric PWA solver available in Hybrid Toolbox?. Simulations are carried out for tracking the vehicle performance for the lateral deviation from the center line of driving lane on a road profile of varying curvature. The paper concludes with comparison of the qualitative performance of the two approaches and the comments on the trade-offs thereon.