Hallo everyone!
I have searched throught the forum, but have not yet seen the same problem as I observed today.

Today I have made a LiDAR mapping survey with DJI Matrice drone together with DJI Zenmuse L1 LiDAR. The survey was conducted in the open field during the good weather conditions. During the survey and after the data were processed I observed several issues:

• Weak RTK signal and RTK signal loses. Matrice 300 was connected to NTRIP. There was a 4G dongle inside the remote control. The Internet connection was about 10 Mbs. The hotspot from a mobile phone was also used. The issue - RTK signal strength on the Remote Control showed no signal strength at all (all poles were grey). Once in several minutes the signal strength got one or more poles, but after several seconds the signal strength marker was grey again. Whenever the Remote Control or just RTK was restarted it took a lot of time for "RTK convergence" - once it took about 30 minutes. During the flight mission connection was dropped several time, mission was paused until the connection was restored. Whenever the connection to RTK existed, position Standard Deviation was about 1-3 cm both horizontal and vertical one.
• After survey was finished and data were processed through DJI Terra I saw that parts of the data are missing. (Shown on the attached screenshot with red color). During the mission drone several time returned to home point, ether when RTK connection was lost (survey was automatically paused) or when the batteries were with low remaining charge (mission was paused after the end of line manually). In all cases the mission was continued afterwards and the drone flighted to the point, where pause was made and continued the survey.
• Beside missed data the rest of the data is corrupted with horizontal and vertical offsets. (See the pictures attached). During the whole survey the standard deviation values on the remote control were monitored - they never jumped out of 1-3 cm diapason. (Beside several times of dropped RTK connection, but the mission was stopped then automatically).
  

In the end, the whole data recorded during the survey day is useless. Just one week before with the same system and the same setup we made a small flight near the office of our company - the result was good and no problems were observed. Since then nothing had changed as far as I know, but this time a lot of problems.

Sounds like the main thing that was different was the location (and thus the network connection quality).

If you have a spotty internet connection, NTRIP doesn't work well. Unfortunately the DJI RTK gear (all of it) doesn't always give the best diagnostics when it is having issues. But I suspect your dongle was not providing a stable enough connection for the system to obtain and maintain a reliable FIX status. Flaky NTRIP will also cause longer time waiting for a FIX. Jumping in & out of FIX is never good for data quality; even when it is not long enough to halt the mission, it still can jack things up a lot. Photogrammetry is much more tolerant of this than LiDAR, due to the sampling frequency (~1/s max vs 80K-240K/s)

As I understand it the EU version of DJI RTK gear will not allow you to use an external hotspot, only a dongle in the RC, which is also unfortunate. But I suggest the following:

• see if you can get a wireless signal quality metering app installed on the RC
• compare other devices withs same & different service providers as the dongle in the location where you had the issues
• try another dongle if possible
• try another survey in a location where you know you have good network connection to establish you haven't developed a problem w/ the dongle or something else
• If you can't get a good connection in the site you need to survey, set up your own base station (D-RTK 2 or better) rather than trying to use NTRIP
  

Good luck!

patiam Posted at 5-17 08:03
Sounds like the main thing that was different was the location (and thus the network connection quality).

If you have a spotty internet connection, NTRIP doesn't work well. Unfortunately the DJI RTK gear (all of it) doesn't always give the best diagnostics when it is having issues. But I suspect your dongle was not providing a stable enough connection for the system to obtain and maintain a reliable FIX status. Flaky NTRIP will also cause longer time waiting for a FIX. Jumping in & out of FIX is never good for data quality; even when it is not long enough to halt the mission, it still can jack things up a lot. Photogrammetry is much more tolerant of this that LiDAR, due to the smapling frequency (~1/s max vs 80K-240K/s)

Technically, I have not seen that it really jumped out of the FIX mode.

Next, hotspot in German version is possible, but did not improved the data anyhow, unfortunately - it was the same "weak" RTK connection as with dongle, while everything worked good on the mobile phone itself.

Last, but not least - problems with RTK is not a reason for the parts of the data missed. At least, as far as I know, terrestrial laser scanner work perfectly without any GNSS connection at all (of course they does not move during the recording, but here there is NO RECORDING). If there is some "very intelligent" algorithm inside of the drone or DJI Terra that just  reject data with bad RTK - then okay, will note it in future. However, as it is possible to make whole missions without RTK (GPS-mode), then it should not be a reason for the data missed.

P.S. DJI support was, of course, contacted as well. If I will receive any meaningful answer from them I will update section here.

Hi @Vermesser. Thanks for the info regarding German version hotspot capability. I do know some countries' versions lack this functionality.

It may be helpful to put an app like NTRIP client on a phone or tablet to assess the stability of the NTRIP data connection in addition to that of the network connection.

I'm well aware that LiDAR scanning can be done w/out GNSS or INS data.  Drone deployment of LiDAR is about the twitchiest platform there is, and a far cry from static scanning. My assumption here has been that DJI's implementation may have some limitations and/or data filtering or QA/QC that interrupts recording or rejects data during periods when RTK is flaky. That may or may not be the case, just spitballing here. You did say that the data that were collected were also compromised. Was there any indication of IMU or INS issues in addition to those with RTK?

I'll be up front that I own neither an M300 nor an L1, but  I do have several Enterprise level DJI drones and have been doing hydrographic and terrestrial mapping using SONAR, LiDAR, photogrammetry, and other techniques for nearly 25 years.

My suggestions were merely ways you might error trap your issues. For all we know, there may be some larger equipment issue here and your initial successful test flight was the exception rather than the rule.

Anyway good luck, and keep us posted if you would!

Hello everyone!

Today I have been at the same spot, where I did the aforementioned survey. Weather was exactly the same - sunny, +25C, open sky, wind 6-7 m/s with the gusts up to 12 m/s.

Things that have changed: all the firmwares (drone, RC, batteries, charging box, L1) were update to the last version of firmware (May update - previous survey was done with the previous firmware); changed mount point of NTRIP RTK corrections - before, as I understood from the corrections provider, the corrections were sent only for GPS and Glonass - now there are corrections for all satellites system that drone receives; RC was connected to the GigaCube (external 4g router with AC connection (to the socket)).

I executed the same missions saved in my RC.

Photos of today results are attached.

Long things short - the results are great! Perfect overlap, just several times RTK outage (with STABLE Internet connection, don't now why. Mission were paused and continued with restored RTK connection).

Problems observed today - once L1 stopped to record data - RC gave an gimbal error. Drone was returned to home, gimbal was checked, no problems were found. I assume, it was because of extreme wind gusts - it happend only once, when a sudden gust of wind raised up to 13 m/s. At this place several additional lines were done afterwards to cover theoretical holes in the data.

One strange hole was found. This hole in the data (with missing data) is located at the another place (2 hours later after the gimbal error) - I did not observed any error during this place survey - on the RC, ar the place of the hole data existed, RTK was fixed (I monitored today it during the whole survey). Another 2 holes exist on the data edges - also don't remember any issues there during the recording.

I assume, this holes are done with dirty DJI Terra hands. - check the attached photos - it looks like the data is cut away with a polygon.

The rest is fine, tomorrow I go this field again and finish it. Hope with no issues. And do the infills, if course. Will make an update soon.

Glad to hear you had better results this time!
Looks like a few variables changed (firmware, internet connection, NTRIP mountpoint & data types) so hard to tell if any (or a combination) fixed the issue, but at least you got good data!

Cheers!