Really, really, freakin' hard.  On subs, the Navy uses dual $20 Million
(pretty sure) gyros to do it, and they are the size of a refrigerator.  To
make a first order approximation of how to do this with an accelerometer-
from basic physics (d=.5*a*t^2)  (It has been a while, so please forgive me
if I did not remember correctly :) if you take an offset of 1/10,000 of a g,
over an hour, the position could be off as much as 6.3 kilometers.  You
don't do a long period INS (intertial navigation system) with cut rate
accelerometers.  You use devices like ring laser gyros and do interferometry
on light. To be within 10m after an hour you need a drift of less than
1.6x10^-7 g.  Not something you're going to get with an off the shelf part.
>From what I've seen, an INS that would satisfy your requirements would be in
the $100,000+ range.

An INS is best over the short term- periodically zeroing out the drift with
some other reference, such as a GPS.  Kalman filters are used a lot to
integrate different sensors to make an accurate INS.  The Ultrasonics, on
the other hand, have a much better chance of success, at least in telling
you range- 3D position is harder unless you have a pretty large array of
transducers to home in on (based on geometric dilution of precision).
Underwater, GPS has its own limitations, unless you are willing to surface
to get a fix on your position.

I believe this topic has been discussed at length before on the PICLIST.  It
is really hard, if you could do it cheaply, many, many people would be
willing to buy it.

Matt Bennett

-----Original Message-----
From: McMeikan, Andrew
Sent: Thursday, May 18, 2000 2:29 AM
To: PICLIST@MITVMA.MIT.EDU
Subject: Re: [OT] inertial guidance for an underwater PIC as a diving
aid
Importance: Low


The subject says it all and if it is feasible we could even get rid of the
[OT] ;)

how hard would this be?  Just need and LED direction indicator to find the
way back to the boat, so accuracy would not have to be better than a few
meters over an hour of swimming (boat would probably drift anyway).

Is this reasonable on a PIC or better suited to an HC11.  Are there reject
accelerometer modules that are not good enough for helicopters (a source of
crashed ones?) but would be OK for diving?  I am thinking cost on this one.

Is there another way (pitot (sp?) tubes??, under water GPS type boat
produced ultrasonics?))

        cya,    Andrew...