Re: [mowbot] Cheap accurate location

Mike Ross (mross nospam at
Mon, 19 May 1997 08:07:28 -0500 (CDT)

On Fri, 16 May 1997, Monta Elkins wrote:

> Unfortunately my other projects are dragging on
> and I haven't been able to focus on mowbot.
> But I have noted the following.
> Delorme makes Tripmate a $150 GPS receiver
> (12 channel parallel?) with a "standard" PC
> serial connection.
> It might be possible to use two of these
> to do differential GPS for positioning.
> I've heard claims of DGPS accuracy to one
> centimeter.
> I'm far from an expert on the subject and
> was looking for other comments....

I'm no expert either, but I tried just what you're suggesting... It
didn't work at all. To do it right, each GPS unit needs to be figuring
off the same satellites, but there's no guarantee that the same satellites
will be visible to both of them. The GPS reading jumps when a satellite
switch occurs, so each unit would have to be synchronized. I'm not saying
it's impossible, but real DGPS with mm accuracy is what, $20K? Doing it
the cheapo way is *amazingly* cheaper, hence that fact that it isn't
being done already tells me there's a barrier. After all, it would be the
perfect way for a mowbot.

I was using units on loan from a Trimble rep, so I only had about 2 weeks
to write and test code. I plotted the computed position of one of the
units, and it would wander around within a 20 ft area, then jump up to 100
ft when a satellite was swapped out.

I also tried averaging. I think if you averaged for a week, you'd get
something pretty close, but you can't do it for moving objects.

* Michael L. Ross/C33 | Lockheed Martin Eng & Sci Svcs Co. ****
* Robotics Department | 2400 Nasa Rd. 1, Houston, TX 77058*****
* (281)333-7094; (281)333-6071 fax* #include <disclaimer.h>*
* Finger me for my public key. mross nospam at ****