DPRG mailing list archives: [DPRG] Further Quandries on GPS...

Wed Jan 2 00:58:18 CST 2008

Kenneth,

Your problem is almost certainly caused by multipath issues.  Have a look
at this link:

http://books.google.com/books?id=peYFZ69HqEsC&pg=PA78&vq=multipath&sig=eceRmyzSTkedaWxeJSIJcTgbO2w

As the GPS satellites move around (they are not geosyncronous) the path of
the signal changes, therefore the reflections change.  The theoretical
maximum deviation from multipath is ~150 meters, but I have seen my
Magellan handheld unit wander over a mile when I was in Finland, in
weather conditions similar to what you describe.

Ed Okerson

> Dean,
>
> I will give the site a thorough read and get back to you, thank you for
> posting.
>
> To fill in some detail as I have run a number of expirements already and
> have verified a few data sets against google earth...   I agree completely
> on the need for more satellites in view.   ProtoBot throws away all data
> until at least 6 satellites are in view.  The data always tracks, meaning
> that the data points reported do not hop around, but rather follow
> transitionary paths.  Often the reported point will stay with-in a few
> feet
> boundary of its last reported position for 15 or 20 minutes at a time,
> however, afterwards it may up and track away 100 or more feet in a matter
> of
> a minute or less.  Often the reported position will return too within
> ~20ft
> of where it should be (google earth reporting) inside of ~5 minutes.
>
> The screen snap shot (sorry for the poor picture) was taken after
> ~20minutes
> of data capture showing a 102 ft by 88 ft wandering path when the unit was
> stationary.   I have captured much longer paths that look similar, however
> my camera has significant difficulty capturing this display w/wo flash
> other
> other settings...
>
> http://nikita.argia.net/kmaxon/sat_wand.jpg
>
> I am most definitely not an expert in the realm of GPS and those that I do
> know who are, work in another division that precludes them from talking
> with
> to me.  Sigh...  So close and yet so far...
>
> -Kenneth
>
> -----Original Message-----
> From: Dean Hall [mailto:dwhall256 at gmail.com]
> Sent: Tuesday, January 01, 2008 11:35 PM
> To: dprglist at dprg.org
> Cc: Kenneth Maxon
> Subject: Re: [DPRG] Further Quandries on GPS...
>
>
> There are plenty variables about your GPS that I don't know about so I
> could
> speculate plenty and that's all it would be. Here is information on
> experiments I have performed:
> http://deanandara.com/Argonaut/Sensors/Gps/
>
>
> Notably, coordinates can be off by as much as 120 meters just after module
> power-up or when in view of only 5 or 6 sats. The SiRFstarIII that is in
> my
> GPS module will give coordinates with increasing accuracy after taking
> multiple readings; a.k.a. after time elapses. Multipath effects are not
> going to be something you can measure (unless you have fancy equipment I
> don't know about; which, if you did, you probably wouldn't be speculating
> like this), so start with what you can measure. I suggest letting your
> receiver sit in one place for a minute or five and examine the behavior of
> the coordinates: are they erratic or are they converging? If they are
> erratic, try again a different location & time. If they are converging,
> find
> out more information, like how many sats in view? and how long until the
> coordinates are within 3 meters of actual (I use Google Earth for that)?
>
>
> By running little experiments like this, I know that my robot warm-up
> procedure will include turning on the GPS module and waiting until a
> significant number of sats have been in view for an amount of time that I
> determine. Only then will I start trusting my GPS coordinate values. I
> have
> written a few Python scripts that help me capture, convert and view data
> from my GPS module. For example, just reading NMEA sentences with the
> receiver at one location for 5 minutes, save the data, convert to a KML
> file, plot it as a path and view it using Google Earth. This helps
> visualize
> where your receiver thinks it is. Scripts are here:
> http://deanandara.com/Argonaut/Sensors/Gps/Scripts.html
>
>
> !!Dean
>
>
>
>
> On Jan 1, 2008, at 21:18 , Kenneth Maxon wrote:
>
>
>   Hello DPRG...
>   Here is one for further conversation in the "Hmmm...?" realm...
>   It was up to a fair 8-deg today and although there is a significant snow
> pack on the ground the sun was bright and the sky was clear for a few
> hours
> so ProtoBot ventured back out into the back yard for more GPS expiremental
> data gathering.
>   This time I made sure that ProtoBot was a full 12 ft away from the house
> and this time, the house does not have a 1/2"~3/4" glaze of ice hanging on
> it. There is however a 3/4" layer of solid ice under the snow back that
> ProtoBot is sitting on. I pulled a bonehead move and forgot to take my
> hand
> held Garmin + adapter & extension cord out to verify the readings,
> however,
> surprisingly ProtoBot still recorded erronous readings, like it was doing
> inside and in the previous outdoor trial up along side the house.
>   I have taken time to swap out the GPS module / antenna combination on
> ProtoBot with the spare unit that I'll be using on an upcoming project.
> And
> given that ProtoBot's last round of testing performed in a similar
> (although
> less drastic from an error standpoint) manner than did the comercial
> Garmin
> unit, I still have a quandry here. Yes, I am still wondering *IF* this
> could
> be coming from multipath off of the ice layer. I theroise that it is not
> as
> the 10" of snow pack sitting on top of the ice should be enough to
> completely disburse the energy, and all of the sattelites reported by the
> module as being used in the actual position calculation had a good signal
> strength reading.
>   I don't know that anyone on this list will have input, however it is an
> interesting quandry that I am working through and I thought it might be of
> interest to the group, as there are others in this group that activly use
> GPS on their robots.
>   Of course I'll post futher when I get to the bottom of the
> issue... -Kenneth
>   BTW, Ron, the package arrived, thank you. You didn't have to return it
> but
> since you did I'll put it to good use in the next project. (IE, you've
> saved
> me from some eBay surfing). -Thanks...
>   _______________________________________________
>   DPRGlist mailing list
>   DPRGlist at dprg.org
>   http://list.dprg.org/mailman/listinfo/dprglist
>
> _______________________________________________
> DPRGlist mailing list
> DPRGlist at dprg.org
> http://list.dprg.org/mailman/listinfo/dprglist
>