Path Tracking of Vehicles on Urban Roads Based on Fuzzy Control and NMPC

Since the curvature of urban roads varies greatly, path tracking in this scenario faces the problem that it is difficult to simultaneously guarantee high accuracy and high efficiency. According to previous work on path tracking, nonlinear model predictive control (NMPC) has good accuracy when the cu...

Full description

Saved in:
Bibliographic Details
Main Authors: Bai, Guoxing, Liu, Li, Meng, Yu, Gu, Qing, Wang, Guodong, Dong, Guoxin
Format: Conference Proceeding
Language:English
Subjects:
Automation
Autonomous vehicles
Control systems
Faces
Fuzzy control
path tracking
predictive control
Real-time systems
Roads
Transportation
urban roads
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Since the curvature of urban roads varies greatly, path tracking in this scenario faces the problem that it is difficult to simultaneously guarantee high accuracy and high efficiency. According to previous work on path tracking, nonlinear model predictive control (NMPC) has good accuracy when the curvature of the reference path varies. But when it is hoped to further improve the efficiency, the accuracy of NMPC-based path tracking control is still difficult to guarantee. So, we combined fuzzy control and NMPC to design a path tracking controller, which can adjust the speed to ensure high accuracy and high efficiency. Then, we verified the proposed controller through the joint simulation of MATLAB/Simulink and CarSim. When the simulation environment is consistent, the maximum absolute values of the displacement error and the heading error of the proposed controller are 80.4 % and 46.6 % smaller than those of the NMPC controller. The proposed controller is robust to positioning error, and it can operate normally when the positioning error is not greater than 0.04 m. Besides, the real-time performance of the proposed controller can be ensured by reducing the control horizon. The maximum calculation time in the simulation only accounts for 28 % of the control period.
ISSN:2688-0938
DOI:10.1109/CAC53003.2021.9727843