Hello, I was just wondering how was the “one best” solution calculated in ES. The option I am referring to is the one available in the settings of the static processing module. Does it averages all the fixed points or does it just select any fixed point at random? How does it work?
Thanks so much in advance!
2 Likes
The final point in static shows the result of convergence of a filter. It is not an average or a random point. It is the best estimate of the position.
6 Likes
Thanks for your answer! Ok, then how can I asses the precision of the point? Is there a way to get the horizontal and vertical precision? What would be the appropriate method for this?
Hi @edgarfer03,
The calculated .pos file with the point contains sdn, sde, sdu, sdne, sdeu, and sdun values. They stand for standard deviations of the solution. You can find the .pos file by tapping Show result files under the Process button.
5 Likes
Could you please elaborate on what that means?
Kalman filters are the core of most GNSS positioning, all scientific & industry standard stuff.
If you want to get an idea on what that means inside Emlid Studio, hang onto your chair and have a look at Appendix E.7 in the RTKLIB Manual where it’s referred to as an EKF:
And a blog here discussing how the filter parameters could be tweaked in RTKLIB settings: Kalman Filter Adjustments in RTKLIB (the Options-Statistics tab) – rtklibexplorer (wordpress.com)
A lot more sophisticated than simplistic averaging or creative artwork.
Otherwise Emlid Studio does a great job of keeping all that bottled up inside and making everything simple for you.
3 Likes
Thanks, the equations are beyond me. Can you or anyone convey what they mean in English?
I think by “creative artwork” you’re referring to my screenshots of the plots. I think it’s valuable to understand the solutions the software is coming up with. I mean those are what is going into the EKF formula right? When you see your fix points scattered wildly and your float points in a tight cluster, that tells you that your starting point is a really complex set of data, and then they’re going into formulas that are in the realm of advanced math.
I’m in a surveying group on Facebook and there was a recent thread on capturing GPS points under foliage. Almost all the surveyors just said they trust GPS under tree canopy because they can get a fix with their hardware and collect a point. It would probably be useful to them to understand what the data looks like by seeing it visually, and then a big picture understanding of what the algorithm is doing with the data.
You know, as opposed to pushing a button.
I found a video by a professor:
Day,
I wanted to confirm @Wombo answer. This point is the best estimate by an EKF (extended Kalman filter). We don’t have articles dedicated to the Emlid Studio algorithms, but there is information about the Kalman filters online, including the video you’ve shared.
1 Like