[OSM-newbies] Will the real track please step forward

digger vermont dv_mlist at verizon.net
Mon Mar 26 04:30:05 BST 2007


Hi Dirk,

Thanks for your response.

On Mon, 2007-03-26 at 01:06 +0200, Dirk-Lüder Kreie wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> digger vermont schrieb:
> 
> > I have a few tracks I've made by walking around. They make me look like
> > a drunken sailor with the amazing ability to right through blocks of
> > buildings! I've another from driving that looks somewhat better.  It
> > actually coincides for the most part with yahoo maps. Is this as good as
> > it gets?
> >
> > I have made some nodes, segments, and ways but at this point I hesitate
> > to upload them because of the accuracy.  
> 
> The SirfStar III Chipset has problems in some situations when you walk
> with it, because it trusts speed measurements more than position
> measurements apparently.
> 

Looks like I didn't give full info on the gps units I have.  The tracks
I walked were with a Magellan Meridian Platinum.  I looked around but
couldn't figure out what chipset it uses.  Is it a SirfStar III?

The track made while driving around was the Haicom HI-303E which has an
Evermore chipset.  I've had the Haicom for several years now and it took
your reply to jog my memory that the HI-303 had several flavors.

Are the comments you made for the Sirfstar III also relevant to these
units?

> Disclaimer: I have no real knowledge about how the SirfStar III does
> stuff internally, this is based solely upon my own observations of the
> behaviour and some knowledge about GPS measurements in general.
> 
> Due to the nature of the sampling inside the GPS receiver it detects the
> diverse movements of the walking body and extrapolates its next
> positions from it, when the positional data is not accurate enough (only
> 3 sats or bad reception) it even "drifts" in that direction for a bit
> longer than strictly needed, thus generating these "drunken sailor" tracks.
> 

The Magellan I would have thought was designed for slow movements such
as hiking.

Could multipath give these type of results?  The houses in the
neighborhood aren't that tall.  Three stories and an attic, about
50ft(15m) 


> You'll find that putting the GPS receiver on your back or shoulder and
> cycling to be much more accurate than walking, because the speed
> measurements and the repeated location samples can work together.
> 

I'm going to experiment some and see what I can see.

Thanks again for your thoughts,

digger






More information about the newbies mailing list