[FOSS-GPS] Inputting a known position.

[Felipe G. Nievinski]

Positional accuracy of satellite imagery is better in relative sense (i.e.,
between nearby pixels) than absolutely (i.e., vs. ground truth GPS), so I
doubt you could get dm-level accurate horizontal positions that way.

RTKLIB allows inputting a fixed position for the rover, but it's
all-or-nothing, whereas you wanted to leave height free. You could try
kinematic positioning with receiver dynamics enabled; its process noise can
be specified separately for each position component, so you could zero the
horizontal one leaving the vertical unchanged (see RTKPOST > Options >
Stats tab).

One last idea is to output solution status / debug trace, extract the
kinematic position estimate (3-by-1 vector, xa) and its 3-by-3 covariance
matrix (Ca); then setup a horizontal constraint, Cc = diag([1e-3 1e-3
1e+3]), xc = [north_known, east_known, 0]; and finally do a weighted
average:
    xp = Cp * (Wa * xa + Wc * xc);
where the final covariance is
    Cp = inv(Wp);
in terms of the weight matrices:
    Wp = Wa + Wc;
    Wa = inv(Ca);
    Wc = inv(Cc);
The a-posteriori height, xp(3), and its uncertainty, sqrt(Cc(3,3)), should
be improved.

-F.

Glenn Thorpe glenn at thorpies.com
Sat Feb 14 14:51:25 PST 2015

> Hello
> Does RTKLIB have the provision to input the fact that I am at a known
> latitude and longitude?
> Nowadays we can often identify our location to within a couple of
> decimeters.  Imagery in Google Earth has pixels of well less than two
> decimeters and the imagery placement seems to be within a few
> centimeters, so if I locate myself on say the end of a road marking I
> know my latitude & longitude to within two decimeters.
> Inputting this known location into the calculation should achieve two
> desirable objectives.  Firstly it will provide an instantaneous
> elevation of within double the error of the latitude & longitude, so in
> the case mentioned an instantaneous elevation of +/- 4 decimeters.  This
> is what many modern pseudo-surveyors want and will really appreciate!
> Secondly it will speed up the time to full fix very considerably, by the
> length of time it currently takes to get to the accuracy of the known
> position.
> Thanks for your time
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/foss-gps/attachments/20150224/4311bece/attachment.html>