DJI Phantom 4 RTK in PPK processing


#21

In this Turtorial it is mentioned to use nasa satellite data


#22

Yep that’s one of the formats, but I don’t know what it is in your systems.


#23

Thanks for your effort, my dropbox says all the data will be uploaded in 6h, maybe this will be helpful that I can publish them. I also received an EZSURV tutorial for UAVs by now but my head is also burning. It would be nice t find a solution tomorrow.

Thanks a lot


#24

PPKRAW.bin should have the NAV info, but needs to be converted in rtkconv for rtklib. Please upload that one. See you later!


#25

Hello,
I agree with HIRSCHRED.

In the Nasa CCDIS data archive you will find the necessary data, above all you need a .nav file.

You can then use it in RTKPOST (as shown in the previous post).

Well, for PPK, the raw data file is especially important. You can convert this into a .obs file with RTKCONV and there is you will find also the .nav file (for GPS).

You use that in RTKPOST. This creates two files: Name.pos and Name_events.pos, this should contain your events. These should match the number of photos.

But I have some questions: You do not use a base station? Where does your correction data come from, from a NTRIPCASTER or D-RTK 2?

I ask because, if you can receive this correction data, then the corrected positions should actually be present in the EVENT file. But you say that you have a difference between the events and the number of photos.

Waiting for your RAWDATA.

Best
But I have some questions: You do not use a base station? Where does your correction data come from, from a NTRIPCASTER or D-RTK 2?
I ask because, if you can receive this correction data, then the corrected positions should actually be present in the EVENT file. But you say that you have a difference between the events and the number of photos.


#26

The D-RTK 2 is a base station and is the only base station that works with the rtk drone. That is just one unfortunate fact.

He already gets the OBS file from the drone. The corrections come from the Potsdam service which is the 18o file you see in rtkpost above. The NAV file is the last one that we were looking for and from the sources that I have it is in the PPKRAW file so that is all we are waiting for.


#27

@MichaelL I thought you were getting the p4rtk? Are you waiting to see how the home grown p4p goes first?


#28

Yes Sir! It’s a band-aid for now as a cheap way to learn the type of system and PPK. The execs know that the proposal is coming. :grin: besides that, I do enough testing so I’m waiting for some of the bugs to get figured out.


#29

Good strategy with DJI products!
I thought you were one of the execs !! :grin:


#30

Sorry my Account is new and I was blocked after 10 replies on the first day (strange thing)
Here is the full data set. It is divided into Jpg pics and all which is not a pic. The pics folder is in 7.2 GB size for 899 items. I think not everyone wants to have this but if somebody wants to perform this as well, here is the opportunity.

Pictures

Not Pictures (the MTK Files are reduced by one empty raw in the end to fit the amount of pics)

The PPKRAW info is very interesting. I will make some trys today

Meanwhile, I was able, with an EZ Surv Rover PPK Tutorial, to PPK the pics but the result is a CSV.
Nevertheless, I merged that with the pics in Agisoft and get a good result in a fast low resolution try.
This night (GMT+1 time - Berlin, Germany) I will perform the generic pics align in highest accuracy and will present you the result here. This means when I am allowed to write again… (what an unnecessary thing for newcomers)
But still, therefore, it is necessary to pay EZsurv and Agisoft as well. An upload to Drone Deploy is still not possible.

And for the question with the correction data, all the work take place without. A station needs a Sim card and without a signal it is useless. Agisoft says the total error is 0.023 m. To check this I would need some GCP. The correction is like shown in the picture here.
image


#31

Haha. I’m usually get what I want, eventually… Just not my pockets.


#32

From my current point of view, I would prefer one of the 3rd party developers with cheaper prices, software and customer support…
I will use the P4 RTK in the end of February in Saudi Arabia and as I have seen, the maximum operating Temperature was reduced from P4 Advance 60degree to P4 RTK 40 degree. Hopefully, it will not melt :wink:
Michael, can you tell me which software you are using currently?


#33

I use drone deploy for planning, flight and processing. Very simple analysis and annotations. Rtklib for post processing if I am using the emlid m+. I use Carlson precision 3D topo to further analyze the point clouds for integration into AutoCAD and navisworks.


#34

Flying a yuneec h520 right now that will be process by drone deploy. No external GPS, yet… I do plan to figure out mounting the emlid reach m+ soon.


#35

This means, you are very familiar with the theoretic workflow but you never had data for a sample like we are trying to execute?
Agisoft went through with the align but not to my sanctification. In the screenshot you can see, that some pics are missing and I will compute this again.
The parameter of the flight were over- and side lapping as default, 80% and 85% and the height 75m. For this the total error of 8cm is bad BUT! There are many trees, snow and a very bad angle of the sun. The site is worst as possible, selected with intention.


#36

Is this the RMSE on the GCP’s or the camera calibration?

I would be happy to take a run at post-processing the images with ACDSee and running it on our account to see how it compares. Just message me a download link and include the GCP file.


#37

The screenshot is showing the RMSE after the camera alignment step in Photoscan. There is another page that shows the 3 components of this error which might be interesting to see.


#38

Thanks. Are the values above GCP’s? Also, did it shoot image between the turns? Not supposed to do that…


#39

Last I used the app in Dec., it captured images in some of the turns.


#40

The values are for the camera position deduced for each of the pictures after camera alignment. They are not specifically for GCP’s. I do not know if the alignment is reflecting the benefit of the GCP’s as I have not processed a mission with GCP’s.

Yes there are pictures included for the small straight line segment between turns. This seems to includes pictures very near the turns. It would be easy to exclude these and re-run the camera calibration.

I would be happy to process the pictures with GCP’s to see if I can find a way to improve the results.