DJI Mavic, Air and Mini Drones
Friendly, Helpful & Knowledgeable Community
Join Us Now

MM Flyaway in Arctic Region

SuperC

New Member
Joined
Jul 24, 2020
Messages
4
Reactions
0
Age
43
Location
Canada
Good day/evening.

I was going to film the vessel/ship I'm sailing on, and ended up having a flyaway.

It was early morning here after shift so I wasn't in my best mental state and egar to fly.

I went on the back deck and started up controller/drone. It asked me to calibrate the compass which worked. For some reason I picked up the drone, and it asked me to calibrate again. Which took 5 tries before it said calibration successful.

I went up in the air with the drone. Then flew it around the back deck with it responding to input. As soon as I flew it off to the side of the ship it took off..

I was putting various control into the sticks to see if I could regain control, at one point I managed to look straight at the ship So started attempting to fly straight to it which seemed to work. Then it started losing attitude, so I gave it up input.. which then I lost control again, and it went sideways into the water.

I did purchase dji refresh and will send in a ticket to see what the can do/think.

This area does have alot of magnetic interference, so I wasn't going to go to far. Concidering the calibration was successful and it flew on the back deck following input I thought it was fairly safe.

Best regards, any input is appreciated.
SuperC
 

Attachments

  • DJIFlightRecord_2020-07-21_[00-03-55].txt
    216.7 KB · Views: 10
I was going to film the vessel/ship I'm sailing on, and ended up having a flyaway.

....

I went on the back deck and started up controller/drone. It asked me to calibrate the compass which worked. For some reason I picked up the drone, and it asked me to calibrate again. Which took 5 tries before it said calibration successful.
Sorry for your loss.

Do you still have a .DAT file on the phone as well as the .TXT file ?
This sounds like the ship was upset the compass and when the drone moved away from the ship where the inertial system said it should be pointing and where the magnetic system said it was pointing didn't line up. When that happens the drone will fly in a curve out of control trying to turn itself to the "right" direction. The data to confirm that would be in the other file.
 
Unfortunately that looks like a simple case of an incorrectly initialized IMU heading due to magnetic interference at start up. The error appears to have been around 90°, which leads to that kind of uncontrolled, curved flight.

Position.png


DJI might honor the refresh on that one given the unambiguous outcome.
 
  • Like
Reactions: BudWalker
It asked me to calibrate the compass which worked.
For some reason I picked up the drone, and it asked me to calibrate again. Which took 5 tries before it said calibration successful.
Concidering the calibration was successful and it flew on the back deck following input I thought it was fairly safe.
That was the alarm going off to warn you that you had a significant problem.
Calibrating the compass was never going to solve the problem because the compass wasn't the problem.
The problem was where you placed the drone to power up and the correct action would have been to switch off the drone and launch from somewhere clear of the magnetic interference.
On a boat, hand launching is much safer than launching from the deck.
What was the deck made of or what steel was very close around or under the drone?
This area does have alot of magnetic interference, so I wasn't going to go to far.
Magnetic influences in the geology won't cause a problem for a drone in the air but a magnetic influence close to the compass at power up can cause serious issues.

Was the vessel moving at the time of startup?
If so, how fast?
 
It reset on takeoff and that moved it by around 800 meters - is that what you are looking at?
At power up it thought it was ~600 metres from home and moving at 3-4 m/s.
The movement might be the vessel.
What was the homepoint it had at the start of the data, when was it recorded?
 
At power up it thought it was ~600 metres from home and moving at 3-4 m/s.
The movement might be the vessel.
What was the homepoint it had at the start of the data, when was it recorded?

Yes - takeoff was at 1.9 s, at which point it was already recording a velocity of 4 m/s to the west. That was probably the velocity of the ship. The initial home point was 800 m to the east, and was reset on takeoff at 2 s.
 
Yes - takeoff was at 1.9 s, at which point it was already recording a velocity of 4 m/s to the west. That was probably the velocity of the ship. The initial home point was 800 m to the east, and was reset on takeoff at 2 s.
But when the data starts the drone is already ~600 metres from the homepoint?
How did this one have a homepoint at 0:00.1 that was 600 metres away?
All flight data I've ever seen has the homepoint being recorded at the start or soon after.
When was the homepoint recorded?
How do you get a homepoint before turning the drone on?
 
But when the data starts the drone is already ~600 metres from the homepoint?
How did it have a homepoint at 0:00.1 that was 600 metres away?
When was the homepoint recorded?

The initial homepoint was in place when the log started, which is at motor start. The original home point was most likely set at power up at that other location, which was about 150 seconds previously if we accept the recorded initial velocity which does extrapolate back to that point.

grab162.jpg
 
Sorry but I don't follow.
When the log started, the data suggests that the drone had already moved 600 metres from where the home point was recorded.
That just does not add up.

Right, but the txt log doesn't begin at power up, it begins at motor start. So if the drone had been powered up 150 s (plus boot time) or so before motor start, then it would have set an initial home point as soon as it had acquired a position lock. Then, 150 s later, when it took off, the home point will have been reset to the new position which was 600 m away, since the ship had moved in that time. The full record would be in the DAT file, since that begins at power up.
 
Right, but the txt log doesn't begin at power up, it begins at motor start. So if the drone had been powered up 150 s (plus boot time) or so before motor start, then it would have set an initial home point as soon as it had acquired a position lock. Then, 150 s later, when it took off, the home point will have been reset to the new position which was 600 m away, since the ship had moved in that time. The full record would be in the DAT file, since that begins at power up.
OK ... He's started the drone and left it a while before starting the app.
Being on a moving ship makes it a little more confusing.
 
OK ... He's started the drone and left it a while before starting the app.
Being on a moving ship makes it a little more confusing.

The app may well have been running before the drone powered up. It just doesn't start recording a txt log until motor start.
 
Sorry for your loss, ships are almost universally made from steel so the compass calibration would have been unusable, and the flight would have ended up with the classic 'toilet bowl' crash.
 
Yes I have attached it.

Sorry for your loss.

Do you still have a .DAT file on the phone as well as the .TXT file ?
This sounds like the ship was upset the compass and when the drone moved away from the ship where the inertial system said it should be pointing and where the magnetic system said it was pointing didn't line up. When that happens the drone will fly in a curve out of control trying to turn itself to the "right" direction. The data to confirm that would be in the other file.
 

Attachments

  • 20-07-21-12-00-01_FLY006.DAT
    2.2 MB · Views: 4
Thanks for your reply.

I had placed the drone on the back deck which is steel, with wood planks.

The ship was moving at approx 6 knots. I was fairly close to the superstructure (accommodations).

I powered up the controller and then the drone. I suppose I was to egar to fly.. and had thought if I lose it, I would only need to replace the drone itself. Kind of regret that decision, after checking later to see it costs more then I thought.

I should had waited until I woke up in the morning to fly. Would had canceled after the failed calibrations, and tried again later (now know to try at a new location).


That was the alarm going off to warn you that you had a significant problem.
Calibrating the compass was never going to solve the problem because the compass wasn't the problem.
The problem was where you placed the drone to power up and the correct action would have been to switch off the drone and launch from somewhere clear of the magnetic interference.
On a boat, hand launching is much safer than launching from the deck.
What was the deck made of or what steel was very close around or under the drone?

Magnetic influences in the geology won't cause a problem for a drone in the air but a magnetic influence close to the compass at power up can cause serious issues.

Was the vessel moving at the time of startup?
If so, how fast?
 
I had placed the drone on the back deck which is steel, with wood planks.

I should had waited until I woke up in the morning to fly. Would had canceled after the failed calibrations, and tried again later (now know to try at a new location).
Launching from on top of or close to steel is something to avoid.
You would need to have your drone several feet above the steel deck to be safe.
Hand launching is the usual method for flying from boats or ships.
 
I'm surprised that the MM did as well as it did. Why didn't it just fly off as soon as it was launched? As noted by @sar104 the IMU Yaw value was incorrect by about 90° at launch and was worse for much of the flight.
1595773279543.png

Even without the problems caused by the ship's magnetic profile the latitude reduces the usefulness of the magnetometer data. The geoInclination (not geoDeclination) is 85° causing the resolution of X and Y magnetometers to be reduced by a factor of ~10. Cos(85 °) =~0.1.

Although visionUsed never turned true I suspect the FC was relying, at least partially, on the vision system to do it's navigating. The vision system was reporting values that matched those shown by the IMU and GPS for the initial part of the flight. The fly away occurred at about the same time (94 secs) the vision system stopped reporting values. I'm not sure when the fly away started, but it seems to be around 86 secs to 96 secs.
1595773823870.png

There were multiple compass calibration requests. The first calibration was requested by the FC because the location was >50 km from the previous calibration
-221.241 : [L-COMPASS]dist_from_last 3236.323 km|

The (1st) calibration appears to have been successful
-173.978 : [L-COMPASS][scale(0)]ok! bias:-1442.5 849.0 1146.7 scal:4.162 5.358 2.168|
-173.978 : [L-COMPASS]app cali success
-173.945 : [L-COMPASS]cali pos&time saved,lat:1.272112,lon:-1.294120,hgt:20.1,dat:20200708

But, then the ship's magnetic distortion caused the "compass calibration required" message and 3 failed calibration attempts were done. Here is the 2nd (1st failed) attempt. The 3rd and 4th are similar
-83.804 : [L-COMPASS][scale cali(0)] estimation error:[19.7]
-83.804 : [L-COMPASS][scale(0)]status: scale exception|

Finally, on the 5th calibration success was achieved. The pilot managed to find a spot where the geoMagnetic field wasn't too distorted. Either that or reduced the physical extent of the calibration dance; i.e. tumbled the MM in his hands as opposed to holding the MM and turning his body.
-12.236 : [L-COMPASS][scale cali(0)] estimation error:[15.8]
-12.236 : [L-COMPASS][scale(0)]ok! bias:-1590.9 679.0 636.3 scal:4.475 4.276 4.433|
-12.235 : [L-COMPASS]app cali success

Instead of re-calibrating after the 1st calibration, which was successful, the MM should have been moved in search of a spot that made the "compass calibration required" message go away. Hand launching is often a good option.

The two successful calibrations don't include the ship's magnetic profile. A calibration only includes the AC's magnetic profile - it can't include any magnetic properties external to the AC. A successful calibration here or any where else would not have altered the outcome of this flight.
 
Last edited:
  • Like
Reactions: sar104
Unfortunately that looks like a simple case of an incorrectly initialized IMU heading due to magnetic interference at start up. The error appears to have been around 90°, which leads to that kind of uncontrolled, curved flight.

View attachment 108800


DJI might honor the refresh on that one given the unambiguous outcome.
Is there really a problem the GPS vs the time integrated IMU locations? The IMU track and the GPS are parallel, more or less. That's usually all we ask for. The initial separation occurs because IMU_ATTI:VelE is held constant while GPS:VelE and IMUCalcs:VelE both are showing the actual movement - about 4 m/s west. IMUCalcs:VelE is time differentiated IMU:Longitude
1595775779464.png
 

DJI Drone Deals

New Threads

Forum statistics

Threads
130,599
Messages
1,554,253
Members
159,603
Latest member
refrigasketscanada