Mavic air - loss of control when connecting to GPS - loss of drone in the sea

[vegardvegard]

Hi there,

I have been browsing through your forums looking for similar cases to mine, and i have decided to post my own experience from flying the drone yesterday evening in an effort to clarify what happened.

It was a calm evening with very little to no wind, and i flew my drone out a couple of meters over open ocean and left it hanging for a couple of seconds. When the drone got connected to GPS signal, it set full speed in one direction, and plummeted into the ocean some distance away. I did not set this movement in motion, and the log shows i did not touch the sticks when this movement started. After it began i obviously tried to stop the movement by attempting to stick the opposite direction and upwards, both without any affect.

Here is the link for the flight: DJI Flight Log Viewer - PhantomHelp.com

I appreciate any help in this matter - i have also contacted DJI to try to figure out what this is.

Thanks in advance,
Vegard

 

[tuscy]

One of the most basic and most frequently repeated pieces of advice is to wait for sufficient GPS satellites AND ensure that homepoint is recorded BEFORE takeoff.

Without GPS the drone will drift and needs an experienced operator to control it.

 

[Meta4]

One of the most basic and most frequently repeated pieces of advice is to wait for sufficient GPS satellites AND ensure that homepoint is recorded BEFORE takeoff.
Without GPS the drone will drift and needs an experienced operator to control it.

That's always good practice but doesn't explain what happened at all.
Drifting or being flown in atti doesn't cause a drone to crash like this one did.

 

[Meta4]

It was a calm evening with very little to no wind, and i flew my drone out a couple of meters over open ocean and left it hanging for a couple of seconds. When the drone got connected to GPS signal, it set full speed in one direction, and plummeted into the ocean some distance away. I did not set this movement in motion, and the log shows i did not touch the sticks when this movement started. After it began i obviously tried to stop the movement by attempting to stick the opposite direction and upwards, both without any affect.

This flight shows some unusual issues which I don't think I've seen before.
Perhaps @sar104 could dig a little deeper if you can post the .dat file from the controller.

You were impatient to launch and flew off in Sport Mode without acquiring any GPS satellites or recording a home point.
GPS and home point happened 38 seconds into the flight when the drone was 50 metres from shore.

The recorded flight data shows full right stick was applied from 0:03.1 until 0:14.1 and the flight mode was switched to Sport Mode at 0:10.3 and to Atti mode at 0:15.5.
The drone picked up speed with the right stick full forward and slowed when the stick was released, coming to rest at 0:19.8.

At 0:32.5 as GPS was starting to come good, the drone started moving again without joystick input, slowly increasing speed.
At 0:35.9 when the drone was flying at 4.3 metres/sec you started to add joystick input, giving it full forward + full right on the right joystick (which accounts for the curved track it flew) and increasing speed.
You also gave full forward on the left stick and some left rudder.
This made the drone climb to about 7 metres above the water.

At 0:42.1 the drone started to become unstable, rolling to one side and pitching sharply forward, causing GPS loss again.
At 0:43.4 you pulled the left stick full back but it probably didn't matter. The drone was already unrecoverable.
At 0:43.4 there are very large changes in pitch and roll possibly indicating a collision.
Any data after this point is suspect.

There were problems with the GPS and warnings about a "weak GPS signal" and GPS positioning off at 8, 11 and 15 seconds.
At 31 seconds it warned again: Abnormal compass function or GPS signal detected. Aircraft switched to ATTI mode.
The drone was warning of a problem and as it turned out, it was not safe to fly.

 

[zenfly2]

My spark always had the announcement "Homepoint has been updated"at takeoff but the Mavic Air I've had for only a week now was silent from day one and I was worried but posted my concerns here I learned that as loong as you see the altitude and distance showing the Air knows where it is. It returns home so it must be true.. I just want to put this here in case anyone else is getting the silent treatment from Miss DJI.

 

[Meta4]

My spark always had the announcement "Homepoint has been updated"at takeoff but the Mavic Air I've had for only a week now was silent from day one and I was worried but posted my concerns here I learned that as loong as you see the altitude and distance showing the Air knows where it is. It returns home so it must be true.. I just want to put this here in case anyone else is getting the silent treatment from Miss DJI.

Not altitude ... just Distance.
Altitude data does not come from GPS which only gives position data which is used to calculate distance from the recorded home point.

 

[waylander]

One of the most basic and most frequently repeated pieces of advice is to wait for sufficient GPS satellites AND ensure that homepoint is recorded BEFORE takeoff.

Without GPS the drone will drift and needs an experienced operator to control it.

What he said...... sorry for your loss though, I always wait for the nice lady to tell me that the home point has been registered, and to check it on the map, makes sense..!
At a new site, I always fly out 50 meters or so in a safe direction, then trigger RTH, when I am satisfied that it is coming back to me I cancel RTH, and go about my mission.

I know that what I said is no good to you now, again my apologies for your loss, but someone else might benefit.

Good Luck
Waylander

 

[zinnware]

Prior to taking off was there a message in the Go4 app indicating a need to calibrate the compass?

 

[Dakarenduro]

Sorry for your loss. What is your flying experience?

I ask because it seems like a common new pilot mistake of being too anxious and jumping the gun. Appears several things attributed to the drone loss... taking off in Sport mode? Applying full stick inputs when in ATTI. Ignoring the warning messages. But most importantly a lack of GPS lock before flying.

Do you have DJI refresh?

 

[Meta4]

To anyone commenting ... read post #4 before suggesting that launching before the home point was recorded.
It was not the cause of the incident.

 

[vegardvegard]

Thank you for all for your input. It is highly appreciated.

The drone did not indicate to calibrate the compass. This indication for calibration happened alot in the start of the drones lifetime when taking off closer to buildings, but not now when i fly more out in the wilderness away from buildings i have not gotten this indication for about 20 flights.

I might have been a bit anxious, but i still feel the drones behaviour after it connected with GPS (at 32.5 seconds) over water was extreme and not controllable, and this being the cause of the accident. I do not have DJI refresh. I have not been able to recove the drone either, so i dont know how much this would have benefitted me.

I will try to get the .dat file from the controller later today (any guide for this?).

 

[sar104]

Thank you for all for your input. It is highly appreciated.

The drone did not indicate to calibrate the compass. This indication for calibration happened alot in the start of the drones lifetime when taking off closer to buildings, but not now when i fly more out in the wilderness away from buildings i have not gotten this indication for about 20 flights.

I might have been a bit anxious, but i still feel the drones behaviour after it connected with GPS (at 32.5 seconds) over water was extreme and not controllable, and this being the cause of the accident. I do not have DJI refresh. I have not been able to recove the drone either, so i dont know how much this would have benefitted me.

I will try to get the .dat file from the controller later today (any guide for this?).

The behavior was somewhat strange. It starts of in P-GPS mode with vision positioning. At 15 seconds it loses vision and, since it still only has 6 satellites locked and a navHealth of zero, drops into ATTI. That works fine, with pitch and roll holding at zero. The FC switched back to P-GPS mode when the satellite count went from 6 to 8, even though the navHealth was only 3/5. That was when the uncontrolled behavior started. Unfortunately it was quickly followed by all kinds of 100% stick inputs, which just exacerbated the problem. The final descent to impact was a result of full down throttle at 43.2 seconds.



The relevant DAT file is on the mobile device, not the controller, and is FLY055.DAT.

Mobile device DAT file: How to retrieve a V3.DAT from the tablet

 

[vegardvegard]

Thanks @sar104 .

The dat file is attached.

 

[CyrusBlack]

What he said...... sorry for your loss though, I always wait for the nice lady to tell me that the home point has been registered, and to check it on the map, makes sense..!
At a new site, I always fly out 50 meters or so in a safe direction, then trigger RTH, when I am satisfied that it is coming back to me I cancel RTH, and go about my mission.

I know that what I said is no good to you now, again my apologies for your loss, but someone else might benefit.

Good Luck
Waylander

Good advise for us newbies

 

[sar104]

Thanks @sar104 .

The dat file is attached.

Well the explanation for the flight behavior is obvious from the DAT file:



There was a 150° discrepancy between the IMU and compass yaw for most of the flight, but the cause is unclear. The aircraft was powered up at -171 seconds, but then there are no further entries in the DAT file data or event stream until around -14 seconds, at which point the IMU yaw and compass yaw disagree as shown. The IMU yaw is normally initialized to the compass yaw, but there is no record of that happening.

I'd argue that's a clear warranty issue since the aircraft appears to be less than one month old.

@BudWalker - any thoughts on that one?

 

[BudWalker]

Well the explanation for the flight behavior is obvious from the DAT file:

View attachment 80593

There was a 150° discrepancy between the IMU and compass yaw for most of the flight, but the cause is unclear. The aircraft was powered up at -171 seconds, but then there are no further entries in the DAT file data or event stream until around -14 seconds, at which point the IMU yaw and compass yaw disagree as shown. The IMU yaw is normally initialized to the compass yaw, but there is no record of that happening.

I'd argue that's a clear warranty issue since the aircraft appears to be less than one month old.

@BudWalker - any thoughts on that one?

It's also possible that the pilot powered up in a geomagnetically distorted location and then moved it to a clean location. This would have had to take place immediately before the .DAT started recording at -14 secs. Otherwise the Yaw/magYaw separation would have been fixed, or at least be smaller than 150°.

BTW, the .DAT didn't start at -171 secs. That's just DatCon using the first tick# and clock speed to determine when the AC was powered up.

 

[sar104]

It's also possible that the pilot powered up in a geomagnetically distorted location and then moved it to a clean location. This would have had to take place immediately before the .DAT started recording at -14 secs. Otherwise the Yaw/magYaw separation would have been fixed, or at least be smaller than 150°.

BTW, the .DAT didn't start at 171 secs. That's just DatCon using the first tick# and clock speed to determine when the AC was powered up.

Of course - so he started the aircraft before the app/RC and the DAT file just references the first tick number. I really hate it when I miss the obvious. That explains the missing DAT data, and I agree that the most likely explanation is that the IMU yaw was initialized with magnetic distortion and the aircraft moved before the DAT started.

And that would change my conclusion about this being a warranty issue - if that's what happened then it becomes pilot error.

 

[zinnware]

I agree that the most likely explanation is that the IMU yaw was initialized with magnetic distortion and the aircraft moved before the DAT started.

And that would change my conclusion about this being a warranty issue - if that's what happened then it becomes pilot error.

sar104, So for clarity to newer pilots, what would the steps be to prevent "Pilot Error"?

 

[sar104]

sar104, So for clarity to newer pilots, what would the steps be to prevent "Pilot Error"?

Firstly - the most important step - check that the aircraft orientation arrow on the map correctly reflects the direction that the aircraft is facing (relative to north). If it is incorrect, as it would have been in this case, uncontrolled flight will generally result. Secondly, don't power up the aircraft and then move it, especially the Mavic 2, which has the capability to notice if the compass yaw diverges from the IMU yaw on takeoff, but not if the divergence happened prior to takeoff. Set the aircraft where you want to take off, then power it up. If you need to move it then restart it after moving.

 

[Splue]

Firstly - the most important step - check that the aircraft orientation arrow on the map correctly reflects the direction that the aircraft is facing (relative to north). If it is incorrect, as it would have been in this case, uncontrolled flight will generally result. Secondly, don't power up the aircraft and then move it, especially the Mavic 2, which has the capability to notice if the compass yaw diverges from the IMU yaw on takeoff, but not if the divergence happened prior to takeoff. Set the aircraft where you want to take off, then power it up. If you need to move it then restart it after moving.

@sar104 Great advice (these are things I wouldn't have ever thought of on my own) and thank you for all the great knowledge you pass along throughout this Forum. As a new pilot with about 10 flights logged I am terrified to watch my $800 pkg leave the ground, especially after reading these horror stories of experienced pilots having random disconnects and flyaways. Youve certainly helped me build a small bit of confidence but I'm still terrified...