Global Positioning System Reference
In-Depth Information
The algorithms used are complex but a common approach is to use variants of
the Kalman filter (see the Epilogue for further details). It uses an estimation
process that is updated based on real measurements and an ongoing motion model.
It can be seen from both this chapter and Chapter 6 that there are many ways
to find position and if Whereness is to be achieved, there must be methods to
usefully combine them. Multisensor data fusion is the key low-level technique
needed to deal with measurements and detailed mapped information (which are
really a priori averaged measurements). The problem at the moment is that there is
a wealth of information that can be sensed, measured, fused, stored, mapped, and
shared but no universal way to fit the “jigsaw puzzle” together. This is a situation
not unlike the more general information maze that is available on the Web. We
know the information is there but its sources and its formats are not yet well
structured. This situation is changing with today's Web 2.0 and the Semantic Web
that is under development. In the next chapter, we will see the fundamental
importance of Web 2.0 methodologies as they are applied to maps and in the final
chapter, we will consider, in the future, how positioning systems could
automatically help determine what can be fused using a Semantic Web approach.
7.8
Summary
This chapter covers a range of different sensor and sensing technologies that are
not radio-based but that can still be very useful to find position. The advantage of
these is that they are generally very simple and cheap solutions and since they do
not use a licensed propagation medium, are license-free. The main disadvantage
from optical and sonic positioning is the need for an unobstructed line of sight.
Optical systems can use either infrared communications usually by simple
proximity or visible light and employ digital cameras. Cameras are being
increasingly used for surveillance and the addition of machine-learning software
means that it is now possible to recognize and track targets within a scene. This is
the outside-in approach. If, however, mobile cameras are used (inside-out), they
can be used by mobile users to capture images of landscapes, skylines, buildings,
and objects that can be compared with a known library of images to provide
positional information. The most accurate positioning systems are optical and are
used for motion capture by cinematic animators. Very complex camera arrays and
software are used to track the detailed motion of key parts of the human body.
Ultrasonic positioning is very accurate (similar to UWB radio). Many
successful research experiments have been performed, but the downside is the
need to install a very high density of sensing units within a building, which can be
expensive.
Inertial and mechanical sensors are useful since they do not need to use any
external medium. Measurement of acceleration by a new generation of MEMS
chips is adding a positioning capability to many consumer appliances. Solid-state
