Is there a tutorial for Antenna Position Compensation and Timedelay Calculation for PPK?

Sorry about the long title, it might help someone else find the answer later if searching.

I’ve got the Mettatech Kit and want to work out my offsets for my antenna and time delays.
Is there are tutorial somewhere for it?

Hi Joe,
could you be more specific about what you are looking for ? Do you have a use case to describe what you mean by antenna position compensation (antenna calibration ?) and timedelay calculation ?

And also, what is the Mettatech Kit ?

Ah apologies.
I’m using a Mettatech PPK unit with M2. When carrying out PPK post processing, you need to work out your antenna offset versus your camera sensor.
Also the timestamp is taken when the LEDs on the Phantom 4 turn off, but apparently there is a time lag and it needs to be accounted for. I’d like to know how to do that as well.
I understand that this is not an Emlid product, but some of the scholars on the forum might know where to look.
Joe

Ok that’s clearer for me now :slight_smile:
But I am not the best to help you on this. Let’s hope someone helps soon enough.

Did you asked the metatec kit seller?

1 Like

I have, nothing as yet.

Are you thinking about how to apply the offsets/lags or how to derive them?

How to derive them

For the derivative of the lags, I would do it the same way as I would derive a calibration:

Make double gridded mission with 4 gcp per 100 images. Lots of overlap, and good surface for reconstruction. Aim for 40 GCP’s or so in a diamond pattern, to make sure you have equal distance from GCP to GCP.
Measure each of the GCP’s using a short baseline, to keep the stdev down.
Now process it all, using GCP as control and camera positions as check marks.
If you plot the horizontal error of each camera position in a histogram (grouped in i.e. 5 cm groups for starters), you should see 1-2 groups peak with the rest following a standard distribution model. Now reprocess with a diff your camera PPK data with a slightly different lag. Rinse, repeat, until you have the largest group at the center of the histogram.

For the Offset, I would suggest just physically measuring it first in XYZ. Doesn’t need to be terrible precise.
When you have your lag figured out, you can apply the same method as above to refine the offset. It won’t be perfect though, as the drone and gimbal angles are not taken into account.
A better way is a total station, or even some of the commercial solutions using markers, preplaced infrared cameras with known positions etc.

2 Likes

Holy cow. Someone must’ve done a video on this, right?
I’ve read what you said (it took a couple of times) and it makes sense, but there are a few gaps in my knowledge:

what am I processing it in? From this point on I have no idea how to use camera positions as check marks.

As a general question: how the heck have we had PPK for so long (>2years) and have so little documentation on it? Have I missed a website or youtube channel?

Joe

I think for your specific case, it’s that if you’re going with non-integrated solutions (i.e. a bolt-on kit that doesn’t communicate with the drone and gimbal), the process is quite involved so many professionals, seeing the hours required as an expenditure, will invest in an integrated system where the problem has already been solved for them.

2 Likes

Indeed, knowing what I know now, I would’ve, but it was a big call back when I knew nothing.
When I have this sorted, I’m writing a manual - its what I do as my ‘main’ income, documentation engineer.

1 Like

No idea, usually I would expect the PPK-trigger hardware provider/manufacturer to at least provide a ballpark lag number for a given number of platforms. I don’t think (or at least naively don’t expect) it changes a lot from model-variance.

I would use Metashape. I believe Pix4D can do it to.
Metashape offers a 30 day full functional trial.

I believe it is still a bit of black art, especially when you have to filter out false triggers etc.

Metta Tech sent me this:

My thinking is it would be 17cm + 36mm to get an average between the L1 and L2?
What do you guys think?

1 Like

Yep… now they just need to tell you lag offset :smiley:

2 Likes

time offset?

theyve come back with 3ms!

Nice! I’ll see if I can apply that to the latest dataset you sent me.

2 Likes

This topic was automatically closed 100 days after the last reply. New replies are no longer allowed.