> I have built a PIC-controlled rover. Now I need to give it
> some kind of moving capability. Does it make any sense to
> try to make it to find its location through a GPS ? I have
> looked in the Net for GPS accuracy and I have seen very
> different values, depending on the technology used. For
> systems using a stationary beacon besides the satellites,
> precisions of cms. can be achieved. Has anybody tested this ?
> Are low level commercial systems (that is, not too expensive
> ones) capable of this ?
>
> If GPS cannot be used, has somebody any idea to know the
> position of my rover (besides IR beacons and triangulation) ?


In practice GPS can give you much better accuracy than the specs seem to
suggest.
Also, the removal last year of the purposefully added "selective
availability" error makes the real accuracy much better than before.

The RELATIVE short to medium term accuracy is well under 1 metre.
That is, if you know your exact starting point then you will be able to
establish your subsequent positions within a metre or perhaps less.

Differential GPS is not especially hard to do. DGPS uses a fixed GPS base
station which know its exact location to provide an error signal that allows
you to correct the error in received GPS information. In addition to the
rover's GPS receiver you need a radio (or similar) link from the base
station GPS to the rover. Some receivers are provided with input modes which
allow a standard DGPS correction "sentence" to be input.
You could almost certainly do your own DGPS. I'm not sure how useful it is
with the removal of selective availability. It will correct for path errors
but I'm not sure how consistent these are between different locations which
are small distances apart.

Using a bare GPS unit I have carried out some measurements of a vehicle
moving around local streets, turning 360 degree loops at intersections,
returning to the same street from a different direction etc. The RELATIVE
accuracy is well under 1 metre. That is, on a relatively narrow residential
road you can clearly tell which side of the road a vehicle is on, and even
where on the road the vehicle is. This is an off the shelf modern domestic
GPS unit - it's the unit used by Rand Macnally (spelling?) with their
automotive map program. Cost is about $US70 AFAIR. It has some degree of
position stabilisation in software but you could probably add this yourself
with software for a unit which didn't have it.

What I am not (yet) aware of is how/if  the relative accuracy changes with
time and whether the position accuracy degrades as distance from the
starting point increases. As I understand the system, neither should change,
but they may.
While you could not complain if the results were ever worse than you hoped
for I feel that the achievable accuracy is liable to be adequate for gross
rover positioning in many cases with added input from local sources such as
wheel movement and steering integration.



        Russell McMahon

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads