Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/85418
Title: Navigation of an autonomous outdoor vehicle on a road surface with changing geometry
Authors: Chan, Fu-man Alick
Degree: M.Phil.
Issue Date: 2000
Abstract: In this study, a truth vehicle model, a reference path model and a road model are developed. Three H∞ control systems, Knom, Kmod and Kblk, are designed based on three design models-1) nominal model (Velnom), 2) multiplicative uncertainty model (Velmod) and 3) system parameters uncertainty model (Velblk)-respectively. The simulations on general real 3-dimensional road surface were conducted. The results show that the H∞ controller (Kblk) developed based on the system parameters uncertainty model has the best performance out of the three controllers. The H∞ controller keeps the lateral off-path distance to less than 20cm and the off-path heading angle to less than 5 degrees, in the presence of system parameters variations, external disturbances and unmodeled dynamics, when a vehicle is running on a path of arc with radius of 183m on a road surface with road bank angle of 5.7 degrees. This research emphasizes the importance of road geometry parameters and shows that the effectiveness of a steering tracking control system is greatly dependent on the road geometry parameters. The present research studies the vehicle control on a general real 3-dimensional road surface. The mathematical model is developed for the dynamics of a four-wheeled vehicle, with front wheel steering, running on a 3-dimensional road surface. The equations of motion are derived are put into state space form to investigate the vehicle-environment system's dynamic characteristics. The truth vehicle model contains 14 states variables. Two design vehicle models (Velmod & Velblk) also involve the steering mechanism. Sensitivity analysis methods and μ-analysis are used to examine the effect of the parameters' variation on vehicle system dynamics and vehicle handling performance. The sensitivity matrices in the frequency domain are calculated as the derivatives of the transfer function matrices with respect to the vehicle-environment system parameters, such as road banking angle and lateral forces. The unmodeled high-order dynamics and the potential modeling imperfection are both taken into account in frequency domain with μ-analysis. A procedure to convert the variations in model parameters onto a structured black matrix is developed in this research. This comes in handy if control engineers wish to use μ-analysis in the development of a robust controller. μ-synthesized optimal H∞ control methods are employed in this research. H∞ controller is designed based on a linearized model for automatic lane changing and path tracking of a vehicle running on a road surface with a constant road bank angle. The yaw angle and lateral distance are the system parameters to be tracked by the steering controller. Satisfactory simulation results are obtained for different banked roads, and a simulation on the general 3-dimensional road surface is also performed. The effectiveness of the controller can be extended to control a vehicle on a general 3-dimensional surface, with time-varying profile. The last Section of Chapter 7 shows that a path on the general 3-dimensional road surface can be constructed in the simulation by joining all the landscapes along the path. With the signals from different system sensors, such as Global Positioning System (GPS), Digital Magnetic Compass (DMC), gyros, accelerator and inclinometer, the actual position and orientation information of the vehicle can be determined on line using the methodology developed in this research.
Subjects: Motor vehicles -- Automatic control
Automobiles -- Motors -- Control systems
Automotive sensors
Hong Kong Polytechnic University -- Dissertations
Pages: 1 v. (various pagings) : ill. (some col.) ; 30 cm
Appears in Collections:Thesis

Show full item record

Page views

40
Last Week
0
Last month
Citations as of Mar 24, 2024

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.