Error Calculator Tool

Galen Rathbun shared this question 1 week ago
Answered

Would it be possible, and is someone interested in, developing a tool or add-on to Locus Maps Pro that would give the user the ability to quantitatively determine the accuracy of an android receiver using Locus Maps?

Here is one possibility: When the tool is implemented, Locus would automatically begin to save points. The each point would be saved at a defined time interval, such as every 15 seconds, and also for a defined length of time, such as 15 minutes. At the end of the length of time, the tool would calculate a convex polygon of all the points and the location of the center of the polygon. Using the average center location, the tool would then calculate the average distance between the center point and each and all of the points defining the polygon. This distance in meters or feet would represent a qualitative error that could be used to compare different Android receivers (phone or tablets) during the same length of time and place, as well as make numerous other comparisons in time and space.

Of course this can be done manually (and is what we biologists do when calibrating error in studies of the use of space by animals), but it would be much easier and more elegant and useful to have a tool to do this.

Comments (7)

photo
1

i´m not absolutely sure what you are looking for ?!

but do you know

https://play.google.com/store/apps/details?id=org.destil.gpsaveraging&hl=en

photo
1

Thanks for the quick reply - and I am not sure what you do not understand ?!  Yes, I am quite familiar with the averaging application.  Perhaps my text did not get successfully attached because the web site was not behaving very well for me.  So, here is what I tried to send:

Would it be possible, and is someone interested in, developing a tool or add-on to Locus Maps Pro that would give the user the ability to quantitatively determine the accuracy of an android receiver using Locus Maps? Here is one possibility: When the tool is implemented, Locus would automatically begin to save points. The each point would be saved at a defined time interval, such as every 15 seconds, and also for a defined length of time, such as 15 minutes. At the end of the length of time, the tool would calculate a convex polygon of all the points and the location of the center of the polygon. Using the average center location, the tool would then calculate the average distance between the center point and each and all of the points defining the polygon. This distance in meters or feet would represent a qualitative error that could be used to compare different Android receivers (phone or tablets) during the same length of time and place, as well as make numerous other comparisons in time and space.

Of course this can be done manually (and is what we biologists do when calibrating error in studies of the use of space by animals), but it would be much easier and more elegant and useful to have a tool to do this.

Let me know if this is not clear and I will try to explain further.  Or, take a look at the methods in this publication: https://peerj.com/articles/1302/

Many thanks, Galen

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Galen B. Rathbun, Ph.D.

Department of Ornithology and Mammalogy

California Academy of Sciences (San Francisco)

c/o P.O. Box 202, Cambria, CA 93428, USA

Office/Home Phone: +1.805.927.3059

Emails: grathbun@gmail.com or grathbun@calacademy.org

www.rathbunX2.com ~ www.sengis.org ~ www.afrotheria.net

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

On 12-Sep-17 12:50 PM, Locus Map wrote:

photo
1

This is very specific request and IMHO it would be, if ever, implemented in a plugin only by somebody who wants it or was persuaded to do it. But the appor plugin developers may think otherwise.


precission = estimation standard deviation, accuracy = estimation vs true value bias. The measurent precission and accuracy is independent.


Such an evaluation is rather about the strength of short term data filtering of static-like GPS data , if applies, i.e. by device GPS firmware( more filtered is not more precise ) and about short term local GPS method precission, determined by device dependent serie of the currently used satelllites sets. Therefore the device comparison may have one place one time validity. The only significant factors I see is the HW sensitivity, used satellite networks ( GPS, GLonass, others ) and ev. static data filtering presence.


For the accuracy, that is influenced by long term or constant factors, one need to know the true position e.g. from precise cartografic data. OSM based map data can have error comparable to GPS location.

photo
1

E.g. a typical pseudorange* standard error is 2-7 m, most of it due variable ionospheric delays. If a device frequently provides deviations significantly smaller, like 1-2m, it is very probable there is implemented time filtering of static-like values, like Kalman one or 2D/3D average or median. The firmware can be intelligent to keep the filtering for static cases and cancel or weaken it for moving states.


Such a filtering obvioulsy helps to determine the value of the estimation, but will not help with the accuracy.


* distance to satellite derived from calculated atomic time and signal timestamp

photo
1

Thanks for the reply and insights.

On 12-Sep-17 11:05 PM, Locus Map wrote:

photo
1

Dear Galen,

I think that Libor already posted quite detailed answer. And I don't really think that such tool would be some useful in Locus app :)

However please check this project/app https://github.com/google/gps-measurement-tools/releases

It seems that from Android N it's possible to obtain raw gnss data and mentioned app can record and prepare some useful stats.

BTW: I guess that would be possible to compare accuracy using observation on some geodetic point that has known location. But as Libor mentioned it's probably needed to perform observation with all device at the same time.

EDIT: following article can be also useful: http://gpsworld.com/innovation-precise-positioning-using-raw-gps-measurements-from-android-smartphones/

Best regards

Petr

photo
1

Dear Petr,

As always, greatly appreciate your excellent software, and also quick responses to address various topics over the years.  Thank you for that web link - I will check it out.

Best, Galen

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Galen B. Rathbun, Ph.D.

Department of Ornithology and Mammalogy

California Academy of Sciences (San Francisco)

c/o P.O. Box 202, Cambria, CA 93428, USA

Office/Home Phone: +1.805.927.3059

Emails: grathbun@gmail.com or grathbun@calacademy.org

www.rathbunX2.com ~ www.sengis.org ~ www.afrotheria.net

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

On 13-Sep-17 12:45 PM, Locus Map wrote: