Page 22 - Autonomous Mobile Robots

6 Autonomous Mobile Robots

1.4.2 Localization and 3D Model Building from Vision .......... 34
1.5 Conclusion ............................................................. 36
Acknowledgments ............................................................ 37
References .................................................................... 37
Biographies ................................................................... 40

1.1 INTRODUCTION
1.1.1 Context
Current efforts in the research and development of visual guidance technology
for autonomous vehicles fit into two major categories: unmanned ground
vehicles (UGVs) and intelligent transport systems (ITSs). UGVs are primarily
concerned with off-road navigation and terrain mapping whereas ITS (or auto-
mated highway systems) research is a much broader area concerned with safer
and more efficient transport in structured or urban settings. The focus of this
chapter is on visual guidance and therefore will not dwell on the definitions of
autonomous vehicles other than to examine how they set the following roles of
vision systems:

• Detection and following of a road
• Detection of obstacles
• Detection and tracking of other vehicles
• Detection and identification of landmarks

These four tasks are relevant to both UGV and ITS applications, although
the environments are quite different. Our experience is in the development
and testing of UGVs and so we concentrate on these specific problems in this
chapter. We refer to achievements in structured settings, such as road-following,
as the underlying principles are similar, and also because they are a good starting
point when facing complexity of autonomy in open terrain.
This introductory section continues with an examination of the expectations
of UGVs as laid out by the Committee on Army Unmanned Ground Vehicle
Technology in its 2002 road map [1]. Next, in Section 1.2, we give an overview
of the key technologies for visual guidance: two-dimensional (2D) passive ima-
gingandactivescanning. Theaimistohighlightthedifferencesbetweenvarious
options with regard to our task-specific requirements. Section 1.3 constitutes

themaincontentofthischapter; here wepresentavisualguidancesystem(VGS)
and its modules for guidance and obstacle detection. Descriptions concentrate
on pragmatic approaches adopted in light of the highly complex and uncer-
tain tasks which stretch the physical limitations of sensory systems. Examples

© 2006 by Taylor & Francis Group, LLC

FRANKL: “dk6033_c001” — 2006/3/31 — 16:42 — page6—#6