Local Positioning System

Unless you use specialized (= $$) equipment, measuring distance with RF
modules will not give you the required accuracy.  The speed of light is
3*10^8 m/s; at 10' ~= 3m, the travel time is only 10 ns.  Not gonna
happen.  You might have better luck measuring angle and triangulating,
but few devices have tracking directional antennas and metal causes RF
signals to bend anyway.

I've seen two systems for indoor navigation, but neither qualifies as
cheap (comercially ~$2k each).

The ultrasonic systems (e.g. Cricket) measure the time of flight of an
ultrasonic chirp, much like ultrasonic rangefinders do.  When the robot
is stationary and you have clear line-of-sight, these systems are
accurate within an inch or two.  My experience with these systems is
that you have low measurement rates, and things mess up when the robot
is moving or nearby objects cause ultrasonic echos.
http://cricket.csail.mit.edu/

Optical systems such as NorthStar identify visual landmarks on the
ceiling in order to navigate.  I've not tried this, but it seems
promising.  Something similar might be done with IR emitters on the
ceiling, a B/W camera on the robot, and optical filters to only pass IR
light.
http://www.evolution.com/products/northstar/devbundle.masn

On the cheap end, I think mapping using dead reckoning and simple
distance sensors is probably the best that's available.

Later,
Daniel

pseudolite technology ( pseudo-satellite )


http://www.google.com/search?hl=en&c2coff=1&q=pseudolite

I had another idea, I know that pinging a server/node is going to be
hard because the speed of Rf waves.  What about AM or changing the
power output of the tranmitter i.e. you lose receiption of radio
station as you drive away.  Some radio station put out 80-100 watts,
could you detect distance with 1-2 watt or even milliwatts, just a
thought.

http://www.fas.org/man/dod-101/sys/land/plrs.htm

google has lots of hits, but I did not see any design specifications.

Google "radio position" or "radio navagition" found the PLRS

I have been pondering IR beacons around my house that would operate at
38.5Khz so the tuned IR receivers would pick them up.  To distinguish one
from the other, each would blink on and off at a differenet rate such as 10
times per second, 20 times per second, etc.

The robot would have an IR receiver on a servo.   The pulse width sent to
control the servo would be carefully calibrated to the angle the servo
turned to at that pulse width so that the robot could know the angle to each
beacon it could see.  The robot would rotate the servo until it got the
strongest signal from that beacon.  Then it would know it was looking
straight at it.

The beacons would have to be near the same level as the sensors on the
robot, presumably near the floor and be omni-directional in the plane of
that level.  If they were mounted high, the robot would have to search a
3-dimensional field to find them and be much more complex.  If mounted high,
they would have the beacons would have to be omni-directional in 3
dimensions instead of just two.

I think a 555 timer at 38.5Khz with another 555 at the on/off frequency used
to turn on/ off the 38.5Khz one would do the trick.  Maybe need to add a
driver to supply enough current to supply a number of LEDs with sufficient
current to make them really bright.  The 555 will supply 200mA.  I think
more will be needed to make the group really bright.

Just my 2 cents worth.