The best results to prevent shortcuts is a GPS system that will sample a
coordinate when your direction threshold crosses a customizable number
of degrees. I've had plenty of succes using about 23 to 26 degrees.
This is not how most handhelds work it seems. I think most of them
work with timers and some sort of algorithm to prevent spidering ( I go
deeper on this subject in this thread :
http://gis.stackexchange.com/questions/35215/detecting-lack-of-movement-on-gps-device
)
I'm using GPS tracers made by KCS (
http://trace.me/index.asp?page=hardware ) , they wil send the data over
GPRS to a backend from where you can extract this data to any format
available.
The results I have with these devices are pretty good. For example:
the cloverleaf in Mechelen (
http://www.openstreetmap.org/way/11186733#map=17/51.05095/4.50413 )
was sampled with 3 units in 1 car, and a Garmin Etrex as control.
I didn't manually / visually correct for mistakes, there where no sat
sources since this was new and google was like 2 years behind bing (who
was a year behind).
But the point of all this is: if you want to avoid the cornering
artifacts: look for a way to sample using heading (degrees) changes.
Other sensors inputs can be : acceleration, distance. Anything that is
not timer based in fact. Timer based samples pretty much suck. A
combination of all should be perfect, as long as timers are only taken
while the asset is driving.
Handhelds compensate (or try to) for this but at the cost of accuracy of
the result set. Sometimes by taking hundreds of samples in a few
seconds, of which plenty will be kind of 'cached' and probably a few
that timeout (in a 2 second timout a car does make a significant
distance ). The the postfilters try to sort those out. It all
depends very much on the implementation quality of the handhelds. My
13 year old garmin etrex still performs quite well given the age.
Hope this helps,
Glenn
On 08-03-14 10:35, Marc Gemis wrote:
The best I get with my Garmin Dakota 10 is 3-4m, but in narrow streets
or forests it might be worse. Another setting on a Dakota is the
number of tracking points. When you don't record enough points you
will notice that the track shows shortcuts on corners.
can't help with the osmtracker questions. Another place to post
questions is http://help.openstreetmap.org , or
http://forum.openstreetmap.org , just in case the Belgian community
cannot help you with this one.
regards
m
On Sat, Mar 8, 2014 at 9:30 AM, <[email protected]
<mailto:[email protected]>> wrote:
I'm still testing a few things, a.o. osmtracker. first headache
was the gps innacuracy. I am using a Samsung S2, and yesterday
osmtracker gave me a 40-50m accuracy, my first tracks looked like
my 3 years old drawing lines. Funny thing is that both google maps
and osmand could show my location with much more precision !?
searching for some web solutions, funny enough, unchecking the
'use sensor aiding' helped a lot. I'm getting 5-20m now. Is that
the best I should expect or am I missing some additional tweaking?
As it seems, osmtracker is a standard in the field, I was
wondering if there is a custom belgian xml file available for the
buttons presets ?
Thanks
Nyamuk
_______________________________________________
Talk-be mailing list
[email protected] <mailto:[email protected]>
https://lists.openstreetmap.org/listinfo/talk-be
_______________________________________________
Talk-be mailing list
[email protected]
https://lists.openstreetmap.org/listinfo/talk-be
--
"Everything is going to be 200 OK."
_______________________________________________
Talk-be mailing list
[email protected]
https://lists.openstreetmap.org/listinfo/talk-be
