A control strategy for improving the accuracy of lateral tracking of autonomous vehicles

Abstract

Lateral tracking is a fundamental technology for autonomous vehicles, particularly in narrow and complex scenes such as parking lots and alleys, where precise control is critical. To enhance the efficacy and precision of the traditional Pure Pursuit algorithm (PP) in the lateral tracking of low-speed autonomous vehicles, this study proposes a continuous adaptive piecewise fitting method based on the cubic Bézier curve to smooth the reference path, followed by a search for the look-ahead point based on the geometric relationship between the vehicle and the reference path. An improved search algorithm based on vehicle speed and road curvature is used to identify the look-ahead point. In case of significant changes in the reference path curvature, the initially obtained look-ahead point is modified. The proposed Improve Pure Pursuit Method (IPP) is then used to calculate the steering angle of the front wheel. To validate the proposed method, the Carsim/Simulink co-simulation model and the autonomous vehicle platform are established. The simulation and real-vehicle tests demonstrate that compared with the pure pursuit algorithm based on Vehicle speed control (VPP) and Fuzzy Logic control pure Pursuit (FPP), IPP performs better in complex low-speed lateral tracking scenes. It reduces the issue of cutting corners when the vehicle enters a corner and enhances lateral tracking accuracy.