Loss of Control

[MML]

Hi folks, I hope someone can give me some peace of mind... I had my first real panic moment today while flying my Mavic.
I was at a work colleague's house this evening, demoing the drone and taking some aerials of his house.

At one point, a couple of moments after takeoff, (in GPS Mode) the Mavic seemed to dart off in one direction on its own, with an apparent loss of control. I had to flick it into Sport Mode to get it back under control.

The location is in fairly remote countryside, the only possible source of interference is a repeater base station for a wireless broadband setup (this model, or one like it: Ubiquiti Networks - NanoStation® M )
Could that have been the cause?

I've uploaded what I think are the correct files from the phone and the drone. I hope these are the relevant ones, but if they aren't please let me know!

Dropbox - DJI_ASSISTANT_EXPORT_FILE_2017-06-05_01-42-05.DAT
Dropbox - DJIFlightRecord_2017-06-04_[17-47-21].txt

For the record the Mavic firmware is .400, the version it shipped with... and the app version is 4.0.7. Phone was on Flight Mode.

Luckily the flight didn't end in disaster, but I'd just like to know what happened, out of curiosity!
Thanks in advance, fellas.

 

[GerdS]

As you have never updated your Mavic, what about compass calibration? Remember, it's thousands of miles away from the factory where it got calibrated last, with total different magnetic behaviour...

 

[MML]

Compass was calibrated before takeoff, actually as part of the demonstration!

 

[BudWalker]

Hi folks, I hope someone can give me some peace of mind... I had my first real panic moment today while flying my Mavic.
I was at a work colleague's house this evening, demoing the drone and taking some aerials of his house.

At one point, a couple of moments after takeoff, (in GPS Mode) the Mavic seemed to dart off in one direction on its own, with an apparent loss of control. I had to flick it into Sport Mode to get it back under control.

The location is in fairly remote countryside, the only possible source of interference is a repeater base station for a wireless broadband setup (this model, or one like it: Ubiquiti Networks - NanoStation® M )
Could that have been the cause?

I've uploaded what I think are the correct files from the phone and the drone. I hope these are the relevant ones, but if they aren't please let me know!

Dropbox - DJI_ASSISTANT_EXPORT_FILE_2017-06-05_01-42-05.DAT
Dropbox - DJIFlightRecord_2017-06-04_[17-47-21].txt

For the record the Mavic firmware is .400, the version it shipped with... and the app version is 4.0.7. Phone was on Flight Mode.

Luckily the flight didn't end in disaster, but I'd just like to know what happened, out of curiosity!
Thanks in advance, fellas.

Interesting. I'm gonna speculate that with the right sequence of control inputs it's possible to confuse the flight controller. Maybe this has been fixed with one of the more recent firmware updates. At time 208 full right aileron had been applied and was abruptly removed. There had also been some left rudder input


This resulted in an abrupt roll from 16 degrees to -16 degrees.


This caused a magYaw/Yaw separation which then led to the FC switching to ATTI mode at time 219 (light blue background). magYaw is computed directly from the magnetometers corrected with roll and pitch. Yaw is the heading prediction made by the FC.

Switching to Sport mode at 224 is shown with the pink background.

Got all that? I don't think this is an incident so much as an illustration that it's possible to overwhelm the FC and cause an error. It would be interesting to see if you can make this happen again. And, then, if you upgrade the firmware see if you can cause this to happen. I'll try this with my Mavic. But it's on the most current firmware, so the comparison may not be all that informative.

 

[BudWalker]

As you have never updated your Mavic, what about compass calibration? Remember, it's thousands of miles away from the factory where it got calibrated last, with total different magnetic behaviour...

IMHO the compass doesn't need to be calibrated because the Mavic has been moved to a new location. The geomagnetic declination isn't determined by a compass calibration. It can't be.

A compass calibration can reveal the strength of the geomagnetic field. But, it appears that the Mavic's compass system is not very sensitive to the overall geomagnetic field strength. It's more robust that the P3 compass system.

I've had my Mavic since October with about 300 flights now. I've never calibrated the compass or the IMU.

 

[MML]

Hi Bud,
wow, thanks for that pretty comprehensive reply! I think I understood you up as far as "Interesting. I'm gonna speculate..."

But on a serious note, what you're saying is, it's likely pilot error as opposed to device error? To be honest I don't even recall what I did, as it all happened so fast. As for the firmware, I left it at .400 on the general advice of friends (if it ain't broke, don't fix it!)

I loaded the data onto Airdata.com last night, hopefully this link works:
Airdata UAV - Flight Data Analysis for Drones

I noticed it mentioned the following:
03m 23s 206.7 ft 106 ft - Data Loss - Downlink data connection lost for 5.7 seconds

What do you think happened there?

 

[BudWalker]

Hi Bud,
wow, thanks for that pretty comprehensive reply! I think I understood you up as far as "Interesting. I'm gonna speculate..."

But on a serious note, what you're saying is, it's likely pilot error as opposed to device error? To be honest I don't even recall what I did, as it all happened so fast. As for the firmware, I left it at .400 on the general advice of friends (if it ain't broke, don't fix it!)

I loaded the data onto Airdata.com last night, hopefully this link works:
Airdata UAV - Flight Data Analysis for Drones

I noticed it mentioned the following:
03m 23s 206.7 ft 106 ft - Data Loss - Downlink data connection lost for 5.7 seconds

What do you think happened there?

No, I think it's a device error. But, it may be a bit unrealistic to expect the Flight Controller to be able to handle those inputs without getting confused. More recent firmwares may work better; I don't actually know. I have noticed that the later firmwares have a tendency to remain in GPS+ATTI and not switch to ATTI when it should switch. Your Mavic did the right thing, IMHO, and switched to ATTI.

Airdata determines the data loss signal by looking at packet arrival times. I.e. they assume the packets should arrive continuously and at a uniform rate. An interruption is considered to be an indication of data loss. Beyond that I don't know much about it. I think this may be documented on their website. Anyway, it's unrelated to the erratic flight.