City University of Hong Kong

CityU Institutional Repository >
3_CityU Electronic Theses and Dissertations >
ETD - Dept. of Manufacturing Engineering and Engineering Management  >
MEEM - Master of Philosophy  >

Please use this identifier to cite or link to this item:

Title: Computational issues on design and implementation of an autonomous guided vehicle
Other Titles: Zi dong dao yin che de she ji yu shi xian zhong de ji suan wen ti
Authors: Makatchev, Maxim
Department: Dept. of Manufacturing Engineering and Engineering Management
Degree: Master of Philosophy
Issue Date: 2001
Publisher: Dept. of Manufacturing Engineering and Engineering Management, City University of Hong Kong
Subjects: Automated guided vehicle systems -- Design and construction
Notes: CityU Call Number: TS191.M34 2001
Includes bibliographical references (leaves 122-147).
Thesis (M.Phil.)--City University of Hong Kong, 2001
xiv, 161 leaves : ill. ; 30 cm.
Type: Thesis
Abstract: This thesis investigates three topics on design and implementation of an autonomous guided vehicle (AGV): (1) dynamic modelling, simulation, and control; (2) algorithms involved in the computer vision system; and (3) humanrobot interface via the Internet. In particular, in all three parts the emphasis is placed on the computational efficiency of the algorithms involved. The problems covered by the first part are related to simulation and efficient trajectory-tracking control of the AGV. The dynamic model utilized in simulation (designed with major contribution from a project collaborator Dr. J. J. McPhee of University of Waterloo) accounts for independent four-wheel steering and driving, dynamics of the DC driving and steering motors and the tire forces that are commonly considered in automotive modelling: longitudinal and lateral friction forces, and aligning torque due to friction. The simulation software is implemented in Matlab in a modular way that allows flexibility in incorporating additional dynamic effects and controllers. The problem of trajectory tracking control for mobile robots with multiple steering and driving wheels has found a number of efficient solutions during last decade. However typical controllers are kinematics-based and do not account for the dynamic effects of a mobile robot. Controllers that are based on sophisticated dynamic models are considered to be computationally demanding and thus often infeasible for real-time processing on an embedded computer. We propose a controller which targets certain dynamic effects of the AGV, namely slippage due to wheel misalignment. Aimed at tracking a piecewise linear approximation of ideal no-slip trajectory in the space of the outplit variables of the wheel plants, the proposed cross-coupling controller is implemented using conventional linear quadratic regulator techniques which implies its respective computational efficiency. The second part proposes a framework to investigate complexity of a class of image processing and pattern recognition algorithms, commonly known as "mesh-local" algorithms (i.e. algorithms that are based on local maps). "Mesh-local" algorithms are extensively used in the tasks that are assigned to the computer vision system of the AGV. We develop a formal representation for such algorithms which allows us to introduce various relevant complexity measures. The operations over the proposed structure are defined and examples of image processing tasks that allow significant reduction of computational complexity are shown. The third part deals with the issues of an agent-based human-robot interface over the Internet. The problems here arose from the specific constraints on the computational capabilities of embedded computers and on communication capacity of the TCP/IP networks. To target both sources of the constraints, a communication language and a proxy-mediated architecture for human-robot interface agents are proposed. The proposed methodologies for minimization of the computational and communication resources required for the AGV-related tasks allowed implementation of the respective control and human-robot interface software using the embedded computer and communication channels of limited capacity thus reducing the final cost of the AGV and broadening the range of its potential applications.
Online Catalog Link:
Appears in Collections:MEEM - Master of Philosophy

Files in This Item:

File Description SizeFormat
fulltext.html161 BHTMLView/Open
abstract.html161 BHTMLView/Open

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


Valid XHTML 1.0!
DSpace Software © 2013 CityU Library - Send feedback to Library Systems
Privacy Policy · Copyright · Disclaimer