Page 167 - Autonomous Mobile Robots

Landmarks and Triangulation in Navigation 151

known locations in the environment; (iii) natural landmarks that are distinctive
features in the environment and can be abstracted by robot sensors; and (iv)
environment models that are built from prior knowledge about the environment
and can be used for matching new sensor observations. Among these envir-
onment features, natural landmark-based navigation is flexible as no explicit
artificial landmarks are needed, but may not function well when landmarks
are sparse and often the environment must be known a priori. Although the
artificial landmark and active beacon approaches are not flexible, the ability to
find landmarks is enhanced and the process of map building is simplified. They
have been widely adopted in many real-world applications, including Global
Positioning Systems (GPSs) [7] and retro-reflective barcode targets [3]. This
chapter only addresses the issues related to artificial landmarks and the associ-
ated navigation methods. More information on other landmarks can be found
in Reference 8.
To make the use of mobile robots in daily deployment feasible, it is neces-
sary to reach a trade-off between costs and benefits. Often, this calls for efficient
landmark detection and triangulation algorithms that can guarantee real-time
performance in the presence of insufficient or conflicting data from differ-
ent types of sensors. Therefore, the use of multiple sensors (laser, sonar, and
vision) and multiple landmarks (artificial and natural) for the position estima-
tion of a mobile robot becomes absolutely necessary. Unlike odometry-based
systems, landmark-based systems do not suffer from drift errors. However,
how to select and recognize good landmarks in different circumstances is a
nontrivial task since the different view angles of landmarks bring different
errors into the measurements. Therefore, it is often the case that some land-
marks are misidentified and this remains a challenging issue in many real-world
applications. Moreover, the cooperative navigation of multiple mobile robots
is a more flexible navigation method than navigation methods for a single
robot.
The rest of the chapter is structured as follows. Section 4.2 presents an
overview of our approach to landmark-based navigation, and proposes a multi-
sensor system that can locate the robot and update different kinds of landmarks
in the robot internal model concurrently. Section 4.3 describes a navigation
system based on a rotating laser scanner and artificial landmarks, in which a
triangulation method for calibrating the mobile robot position is also presented.
Then the visual-based navigation is addressed in Section 4.4 for the mobile robot
torecognizethedigitalandsymboliclandmarksautomatically. Theselandmarks
are very common in office environments (name plates) and highway systems
(road sign boards). Section 4.5 describes the localization system based on a
SICK laser scanner and two cylinder landmarks, in which cylinder landmarks
arefixedontwomobilerobotsandcanchangetheirrelativepositionanddistance
for localization. Finally, a brief summary and potential future extension are
given in Section 4.6.

© 2006 by Taylor & Francis Group, LLC

FRANKL: “dk6033_c004” — 2006/3/31 — 16:42 — page 151 — #3