PILOT 2 CALCULATING WRONG AGL DATA ( Follow terrain feature )
< ! >  To all PILOT 2 users, you might encounter that beug as well, ad we tried it on 5 diffrent remotes ( M300, M3E, and M3T ) all share the same bug whatever version. ( Even on the last update )




Description of the bug ( Altitudes are simplified to simplify the understanding )
For the next part, we assure that 100% of the DSM were in ELLIPSOID, including the one downloaded from DJI server of course. This is not a GEOID error.


- Step 1 : Importing of drawing a corridor route (VOL1.kmz )
- Step 2 : Choosing some basic flight planning ( side lap / band extension etc - this part doesnt matter )
- Setp 3 : Activating the "follow terrain" feature and load up 4 different DSM

• DJI Autoimproted dem.
• DEM_1M : DSM with 1m / pixel of all the valey
• DEM_half_1m : DSM with 1m/pixel of half of the valey
• DEM_HR : DSM with just the are concerned by the flight with 30cm / pixel
  

( I rename my files DEM, but they are in reality DSM files - You can open it with agisoft for exmaple, it will show trees correctlly )


At the bridge ( lowest part of the flight ) ALL DEM, including DjiFligt HUB show an elevation of +816m ( Above Ellipsoidal height )
Both Metashape and Flight Hub read the same HAE ( approx 820m )
Metashape :

Flight hub :




I'm planning a flight route at +56m AGL and tried for all 4 different DSM files ( including the one provided bu DJI )

DJI DSM :


HR DSM :



DJI PILOT 2 nows calculate an approximative flight height of 828m which is only +- 10m above my ground level. Instead of the +56 asked.


We did that on all 5 remotes, + some are tuning the latest version available for pilot @, all show the beug.


< ! >  WHEN FACTORY RESTING the remote controler, and planning the same mission, THE AGL ALTITUDE CALCULATED is now CORRECT.
BUT : Turning off your remorte, and turning it back on makes the error appears again.
We managed to do the flights with another flight plannig app.

We tried to launch a mission that had a similar bug and it appears that the dorne went to the CORRECT altitude. It appears for us that DJI PILOT @ is just displaying WRONG DATA, but still sending CORRECT data to the drone ( Heigt AGL )


Please let us know if you have the same bug on your RC, here is a dropbox link with all files (KMZ, DSM, even if you try with the DJI DSM the bug still occurs on that particular AREA )


https://www.dropbox.com/scl/fo/x ... z1adsx0u9c&dl=0

I get the same result using your rasters.

However, this is a terrain-follow function, not a structure scan. Your rasters are chaotic (DSM). The terrain follow must be done on the ground (DTM), not on the vegetation because it is absolutely not a collision avoidance system..
You will certainly get better results using elevation data such as SRTM-3. Shuttle Radar Topography Mission (nasa.gov)
or else European Digital Terrain Models (EU DTM) - OpenGeoHub Foundation: Connect | Create | Share | Repeat
At least I never had any altitude problems using DTMs. Maybe a stroke of luck!

If you want to focus your study on the canopy, it is recommended to either reduce the resolution of the DSM raster to make it "smooth", or correct the target altitude of the DTM.

Afterwards, if you really want to monitor DSM with the precision of the shape of the trees, you need more efficient flight planning software (UGCS).



And for the purists, this does not affect the problem mentioned but it gives me the itch so I say it anyway, a WGS raster cannot be in meter, but in arcsecond !



good luck

"even if you try with the DJI DSM the bug still occurs on that particular AREA"


Just seeing DJI + DSM makes me say, DO NOT USE!

LV_Forestry Posted at 12-6 00:57
I get the same result using your rasters.

However, this is a terrain-follow function, not a structure scan. Your rasters are chaotic (DSM). The terrain follow must be done on the ground (DTM), not on the vegetation because it is absolutely not a collision avoidance system..

Hi,
Thanks for your reply and YES i'm using it as an anti collision system !
But you're right, if that's the ground i was looking for, i would havegenerate a DTM

LIX Posted at 12-6 07:56
Hi,
Thanks for your reply and YES i'm using it as an anti collision system !
But you're right, if that's the ground i was looking for, i would havegenerate a DTM

For collision avoidance there is the CSM radar from DJI. No idea what it's worth, never tested.
https://www.studiosport.fr/radar ... rtk-dji-a21398.html

Otherwise I recommend that you contact SHP engineering, they have really effective solutions that I have been able to test.
https://www.sphengineering.com/integrated-systems

LV_Forestry Posted at 12-6 09:30
For collision avoidance there is the CSM radar from DJI. No idea what it's worth, never tested.
https://www.studiosport.fr/radar-csm-pour-dji-matrice-300-rtk-dji-a21398.html

Hi !
Yes, thanks for your reply !
We also use DJI ME3 a lot for these missions, and unfortunately there's no CSM available for it ... So we need to have a good DSM / DTM software to rely on.

Trees here can be > 25m sometimes...

But in that case, we're not talking about the topic, the topic is that DJI Pilot 2 is sending my drone right to the terrain if I look at what the height shows ... And that's my major concern ... :/

LIX Posted at 12-7 00:21
Hi !
Yes, thanks for your reply !
We also use DJI ME3 a lot for these missions, and unfortunately there's no CSM available for it ... So we need to have a good DSM / DTM software to rely on.

Yes i believe its because the raster has sudden change of altitude. Have you tried with DTM ?

LV_Forestry Posted at 12-7 00:38
Yes i believe its because the raster has sudden change of altitude. Have you tried with DTM ?

Not yet, but what's weird is that with a defaut factory reset controller everything looks fine, and as soon as you reboot it, the altitude changes ... That's the major bug i guess...
Why would the altitude change when you just turn your controler in sleep mode ? ...

LIX Posted at 12-7 06:29
Not yet, but what's weird is that with a defaut factory reset controller everything looks fine, and as soon as you reboot it, the altitude changes ... That's the major bug i guess...
Why would the altitude change when you just turn your controler in sleep mode ? ...

On that point i have no clue. Some mystical programming issiue.

Anyone have the same bug here ?