Page 224 - Introduction to Autonomous Mobile Robots

Mobile Robot Localization

Figure 5.19 209
An artificial landmark used by Chips during autonomous docking.

the potential to be both compact like a topological representation and precise as with all
direct geometric representations.
Yet, a chief motivation of the topological approach is that the environment may contain
important nongeometric features – features that have no ranging relevance but are useful
for localization. In chapter 4 we described such whole-image vision-based features.
In contrast to these whole-image feature extractors, often spatially localized landmarks
are artificially placed in an environment to impose a particular visual-topological connec-
tivity upon the environment. In effect, the artificial landmark can impose artificial struc-
ture. Examples of working systems operating with this landmark-based strategy have also
demonstrated success. Latombe’s landmark-based navigation research [99] has been
implemented on real-world indoor mobile robots that employ paper landmarks attached to
the ceiling as the locally observable features. Chips, the museum robot, is another robot that
uses man-made landmarks to obviate the localization problem. In this case, a bright pink
square serves as a landmark with dimensions and color signature that would be hard to acci-
dentally reproduce in a museum environment [118]. One such museum landmark is shown
in figure 5.19.
In summary, range is clearly not the only measurable and useful environmental value for
a mobile robot. This is particularly true with the advent of color vision, as well as laser