Motion Vision: Design of Compact Motion Sensing Solutions for Autonomous Systems Navigation

There are no files associated with this record.

Title Motion Vision: Design of Compact Motion Sensing Solutions for Autonomous Systems Navigation
Author Kolodko, Julian Paul; Vlacic, Ljubo
Year Published 2005
Place of publication London, United kingdom
Publisher The Institution of Electrical Engineers (subsequently the Institution of Engineering and Technology)
Abstract This research monograph focuses on the design, development and implementation of a motion measuring sensor. Because it is difficult to fit the description of a design into a traditional textbook style, we have opted to present our work in more of a project style, beginning with background information before moving onto the design and test of our algorithm and the implementation of that algorithm in hardware. Our project oriented style makes this book more informational than tutorial  extensive references are given for those readers wishing to further research a particular topic. We begin the book with an introduction to estimation theory  the fundamental concept behind our work. Based on this we give a broad introduction to the idea of motion estimation. Most attention is given to the gradient based techniques used in our sensor however we present all main motion estimation techniques. From here we develop and analyse our algorithm and its components in detail using simulation tools. The inspiration for this book stems from our work in developing control systems for cooperative autonomous vehicles. While implementation of sensing systems for static obstacle avoidance was relatively straight forward, these systems did not directly allow a very simple form of interaction: intelligent avoidance of moving objects (i.e. other vehicles). How were the vehicles to make intelligent navigation decisions in a dynamic environment without a sensor providing environmental motion data? In response to this need, we developed a prototype intelligent motion sensor. This sensor was implemented as a system on a chip, using Field Programmable Gate Array (FPGA) technology and it uses a combination of visual information (from a camera) and range information (from a scanning laser range finder) to give an estimate of the motion in the environment. The process of building this sensor lead us through a range of fields from robust statistics and motion estimation to digital design. As we progressed, we developed a number of novel techniques that seemed to have value to the broader estimation and motion estimation community. It also became apparent that our design process would serve as a useful case study for those with an interest in digital system design by providing direct illustration of practical design issues. As such, this book is primarily intended for those interested in real-time motion measuring sensor design where the sensor does not simply estimate motion in pixels per frame, but gives a measurement of motion in (centi)meters per second. Also, it is aimed at assiting undergraduate/postgraduate students and graduate engineers who have a grasp of programming & basic digital logic and are interested in learning about programmable logic devices  FPGAs in particular. We present the design of our digital motion measurement sensor in a way that will make the leap from programming PC's to programming FPGA's easier for those who already have a grasp of standard programming techniques. Furthermore, we present the sensor design from a project-oriented perspective rather than from a more traditional textbook perspective. We believe this project oriented approach gives a more solid grounding since the reader immediately sees the relevance of design issues  these issues do not remain theoretical as they so often do in textbooks. Our review of estimation & motion estimation also makes this book a valuable introduction to the concepts of motion processing for those in fields of computer vision and image processing motion. Finally, our presentation of a number of novel techniques (the LTSV estimation, dynamic scale space, use of range to segment motion) makes this book valuable for researchers in the fields of estimation and motion estimation. This book is structured to lead you through our design process. Introductory material on the topics of estimation theory (chapter 2), motion estimation (chapter 3) and digital design (
Peer Reviewed Yes
Published Yes
Volume 67
Edition 1
ISBN 0863414532
Date Accessioned 2006-03-10
Language en_AU
Faculty Faculty of Engineering and Information Technology
Subject Interdisciplinary Engineering
Publication Type Books (Authored Research)
Publication Type Code a1

Show simple item record

Griffith University copyright notice