People don't typically think about places in terms of coordinate locations. Instead,
people tend to think about places mainly in terms of what things are at places, what
activities occur in places, and where places are relative to other places of the
same sort (rooms, buildings, blocks, cities, states, countries, etc.). Airline pilots
and ship captains understand the importance of location coordinates in navigation,
but most of us get to places using addresses, maps, and transportation services.
Besides location coordinates, places have a spatial and temporal extent (the inside of a
car while on a trip), though usually we think of places with relatively fixed locations
that can be recorded on maps and floor plans. This section deals with the problem
of determining the location of a fixed place on Earth in terms of its coordinates.
Location on Earth is specified in terms of longitude, latitude, and elevation. WorldBoard
requires determining location to an accuracy of under one meter. Finer and coarser
divisions are possible, but the cubic meter is the preferred basis for WorldBoard. There are about 10**14 (one hundred million million) square meters on the surface
of the earth. Since people travel in the skies, underwater, and underground, there
are about 10*20 (one hundred billion billion) cubic meters of space around the planet where a person might be (excluding space travel).
For determining location, a combination of GPS (Global Positioning System)[5,6], DGPS
(Differential GPS)[7], and INS (Inertial Navigation System)[8] technologies can be
used. Currently, GPS technology is only accurate to within 10 meters outside, while
DGPS technology can be used indoors or outdoors with sub-centimeter accuracies, but
at a much greater cost. GPS requires satellites, and DGPS requires both satellites
and radio beacon "pseudollites." INS systems are mostly based on accelerometers
(devices for measuring the acceleration of an object). A key application of accelerometers
is in cars for air bag deployment, so the cost of accelerometers is being driven
down enormously in recent years. Although some companies that produce accelerometers
see their main business as air bag deployment sensors, many already have an eye on virtual
reality applications [9].
INS accumulate errors [10]. Accelerometers provide information about acceleration
and by integrating twice, position can be estimated (acceleration * time = velocity,
velocity * time = distance). Each integration step adds errors, and without resetting
the errors eventually become so large that the position estimate is no longer accurate.
INS systems in cars solve the resetting problem by relying on turns in the road
with known positions, so that the position can be reset each time the car turns.
WorldBoard could make use of this technique if a map or other position model is available.
Alternatively, when the client goes outdoors the GPS system can do a reset. When
the client is indoors either the INS can be used, or if the space is equipped with
a DGPS system DGPS technology can be used. Statistical techniques can be used to refine
the accuracy of GPS as well as perform resetting without a priori maps.
While DGPS is most accurate today, ultimately INS will become the preferred technology
to use in WorldBoard. Unlike DGPS which requires costly infrastructure, satellites
and radio beacon pseudollites, INS can be a completely closed system that is small,
light weight, low power, and eventually very accurate. It is interesting to consider
the development of the first chronometers or clocks for accurately measuring time.
When watches were first invented they were not accurate for very long periods and
had to be reset often [11]. Like current INS systems, early watches accumulated errors.
Today's watches form a highly accurate planetary information system. As INS technology
improves, INS devices (or geometers) will be set in the factory and operate for years providing sub-centimeter accurate position information. To operate for one year
accurately, with no resets, if the INS device is doing a double integration to determine
position, then the acceleration error would have to be less than 1 part in 10**14.
Lest these seem like an impossibly high accuracy rate, experiments are currently
underway that measure acceleration based on the phase shift of light. Just as atomic
clocks are necessary for extremely accurate time measurements, atomic level phenomena
will probably be needed for extremely accurate position measurements. In addition,
there are techniques for resetting based on recurring patterns of behavior, magnetic
field anomalies, etc. Also, just as we reset our chronometers when we go to new
time zones, it is not inconceivable that we will reset our geometers when we start out in
a new city. Radio beacons in airports could provide this information.