Help me figure out why the M2P crashed ?

[Dakrisht]

So after 200 flights and 1,000,000+ feet flown the inevitable happened - the M2P suffered a fatal crash this evening.

Replacement is on the way but I’m very curious to find out what happened and how to prevent this from happening in the future. I know there are some all-star black box diagnositicians on the board so curious to here your analysis.

Currently reading up on how to pull flight data from the app / drone... but here’s the play by play.

1) Take off was from a sidewalk, M2P initialized, initial compass error. Fixed after calibrating.

2) GPS took some time but did initialize and recorded the home point.

3) Take off, hover to 30-40 feet, quick joystick to the right to clear trees above.

4) Suddenly, M2P starts flying backwards, uncontrollably, full speed, veers right, and slams directly into a building wall. Dead on impact.

Note: the compass error was present upon initial boot up. I’ve seen the compass error many times before, but never experienced anything like this obviously.

I’m aware of magnetic interference coming from tall buildings or buildings in general. But have flown successfully many times when taking off from a sidewalk near tall buildings.

All collision systems on the M2P were active.

Anyhow, look forward to diagnosing.

I’m thinking compass error caused it to veer uncontrollably. No input on the control could save it.

 

[WithTheBirds]

I’m the abscence if the flight log being available for analysis I will have a crack.

The reported compass error was attributed to a bad take off location. The calibration you performed succeeded in accounting for the local anomaly in the magnetic field. As soon as you flew away from the disrupted magnetic field the compass reading would have been off.

I have managed to replicate this with my old P3. Calibrated over a reinforced concrete cricket pitch. As soon as I flew away from the ground the compass error would trigger and switch to atti mode. The heading depicted in the app was 180 deg out from actual. Bringing it in to land the right stick was basically reversed.

 

[Dakrisht]

Here's the data I just pulled from the Go4 app... Ran the data in the AirData tool which shows compass as OK. But very strange behavior.

Take off location was not optimal, most likely did not think twice about this and I should have. Weather conditions were perfect, no wind, visibility good, plenty of sky for GPS above, wide 4-lane street separating both buildings...

Would a compass error prevent the bird from hovering in place vs. the uncontrollable full-speed flight backwards uncontrollably?

Data attached!

Adding: Bird never switched to ATTI mode, and if it did, ATTI should have kept it at level altitude via the barometric sensor. Why it suddenly flew back at full speed, without joystick input after hover and the quick position change 10 feet to the right (to clear trees) and crashed into a building is the mystery to me...

Also adding: GPS data per the Airdata site is not accurate. I've added the actual location to the image attached... Strange... GPS was good, 9-10 satellites, and green on the Go4 app.

 

[Meta4]

Here's the data I just pulled from the Go4 app... Ran the data in the AirData tool which shows compass as OK. But very strange behavior.

Your whole flight was done with GPS health of 0-1 (out of 5)
The data you posted doesn't show the crash.
It's a 24 second flight and the data just stops with the drone 47 feet up and flying at 18 mph.

DJI Flight Log Viewer | Phantom Help

Upload and view flight logs from your DJI drones

www.phantomhelp.com

It looks like you lost connection before the interesting part started.

1) Take off was from a sidewalk, M2P initialized, initial compass error. Fixed after calibrating.

Fixed??
Your compass was warning you of a local magnetic field problem.
Re calibrating your compass couldn't possible "fix" the problem the compass warned you about.

4) Suddenly, M2P starts flying backwards, uncontrollably, full speed, veers right, and slams directly into a building wall. Dead on impact.
Note: the compass error was present upon initial boot up. I’ve seen the compass error many times before, but never experienced anything like this obviously.
I’m aware of magnetic interference coming from tall buildings or buildings in general. But have flown successfully many times when taking off from a sidewalk near tall buildings.
I’m thinking compass error caused it to veer uncontrollably. No input on the control could save it.

The symptoms you describe are classic Yaw Error symptoms.
many similar stories have been reported from launching from reinforced concrete surfaces.
Launching from reinforced concrete surfaces is just asking for trouble.
It had nothing to do with the buildings and everything to do with initialising your mavic in a magnetic field, probably caused by steel reinforcing in the concrete sidewalk.

 

[WithTheBirds]

If you could have launched without calibrating it or hand launched (a few feet makes all the difference) you probably would not have had an issue.

 

[Dakrisht]

Your whole flight was done with GPS health of 0-1 (out of 5)
The data you posted doesn't show the crash.
It's a 24 second flight and the data just stops with the drone 47 feet up and flying at 18 mph.

DJI Flight Log Viewer | Phantom Help

Upload and view flight logs from your DJI drones

www.phantomhelp.com

It looks like you lost connection before the interesting part started.
Fixed??
Your compass was warning you of a local magnetic field problem.
Re calibrating your compass couldn't possible "fix" the problem the compass warned you about.

The symptoms you describe are classic Yaw Error symptoms.
many similar stories have been reported from launching from reinforced concrete surfaces.
Launching from reinforced concrete surfaces is just asking for trouble.
It had nothing to do with the buildings and everything to do with initialising your mavic in a magnetic field, probably caused by steel reinforcing in the concrete sidewalk.

Connected all the way up until the crash. The battery completely snapped off the drone post-impact. The last data point is most likely the battery snapping off the body.

I guess definitely not fixed in this case. Assuming the compass was re-calibrated in a place with magnetic interference due to a concrete reinforced sidewalk - obviously something to seriously avoid in the future - would this explain the sudden and rapid yaw movements rendering the drone uncontrollable?

Trying to understand why it 1) didn’t switch to ATTI in order to hover or 2) simply stay in one place without deciding to fly by itself at full speed into a building. And seems all of the collision systems didn’t do anything either as it went into said building either.

 

[Dakrisht]

If you could have launched without calibrating it or hand launched (a few feet makes all the difference) you probably would not have had an issue.

Hand launch seems like a better idea in retrospect. Was not aware of the concrete reinforced sidewalks causing major magnetic interference compass calibration or not.

Just trying to wrap my head around what to do when magnetic interference does occur, especially in flight, in order to avoid crashes or simply to hover and land in a safe place. The rapid, full speed, backwards yaw makes no sense. Yet.

 

[Dakrisht]

Something I seemed to have missed:

“The magnetic interference message is prompting you to move to a different location (not calibrate the compass).”

Good $1600 lesson. But thanks for insurance.

 

[Deleted member 94047]

Assuming one had the situational awareness and the time to react, would the "pause" button stop the drone dead on its tracks in these situations?

 

[Meta4]

Assuming one had the situational awareness and the time to react, would the "pause" button stop the drone dead on its tracks in these situations?

No .. the drone was completely out of control just trying to maintain position.
Pause button wouldn't work.

 

[Meta4]

The last data point is most likely the battery snapping off the body.

If the crash was recorded, we'd see it in sudden large changes in the pitch, roll and yaw data, but the data stops with the drone in a stable flight.

I guess definitely not fixed in this case. Assuming the compass was re-calibrated in a place with magnetic interference due to a concrete reinforced sidewalk - obviously something to seriously avoid in the future - would this explain the sudden and rapid yaw movements rendering the drone uncontrollable?

The calibration was probably not the issue.
Starting up within the magnetic field was the cause.
At startup the drone is setting the sensors to match up with what should be the earth's magnetic field.
The difference between the alignment of the problem magnetic field vs the earth's magnetic field affects how badly the affected drone will perform.

Trying to understand why it 1) didn’t switch to ATTI in order to hover or 2) simply stay in one place without deciding to fly by itself at full speed into a building. And seems all of the collision systems didn’t do anything either as it went into said building either.

Once the drone's sensors are screwed up it can't stop and hover.
It was chasing it's tail trying to correct itself but the sensors were giving it wrong data so the effort to correct just made things worse.

 

[Meta4]

Just trying to wrap my head around what to do when magnetic interference does occur, especially in flight, in order to avoid crashes or simply to hover and land in a safe place.

You saw how it behaved.
Prevention is the only real way to deal with this potential hazard.
Avoiding launching from reinforced concrete or steel surfaces would avoid most incidents like this.
You can also check the radar display before launching to see that the drone is pointing the right way according to the radar display and the north point on the display is aligned properly.

 

[sar104]

As @Meta4 concluded, this was almost certainly due to a distorted magnetic field at the takeoff point. As it climbed out of the distorted field the IMU and magnetic yaw values diverged, and that's reflected in the disagreement between the IMU velocity and position.

Just trying to wrap my head around what to do when magnetic interference does occur, especially in flight, in order to avoid crashes or simply to hover and land in a safe place. The rapid, full speed, backwards yaw makes no sense. Yet.

It's the result of an unstable positive feedback loop in position control. Consider the situation where, as a result of a yaw error, the aircraft is facing north but the FC thinks it is facing south. Now imagine that a slight breeze from the south starts to push the aircraft to the north. The FC detects that motion to the north and knows that it needs to exert thrust to hold position. But, because it thinks it is facing south, it applies forward thrust (negative pitch). It's actually facing north so that exacerbates the drift to the north, and so it applies more forward thrust to compensate. Now you have uncontrolled linear flight to the north.

 

[Deleted member 94047]

This was posted by a member a while ago, I found it immensely helpful in understanding magnetic interference and, more importantly, why one shouldn't re-calibrate the compass while within range of the magnetic interference. OP, if you haven't watched it before, you may find it helpful.

The thread i took that link from is here

Signal interference

What things/situations/etc. will interfere with the signal between my Mavic Pro and the controller? Thanks.

mavicpilots.com

 

[sar104]

This was posted by a member a while ago, I found it immensely helpful in understanding magnetic interference and, more importantly, why one shouldn't re-calibrate the compass while within range of the magnetic interference. OP, if you haven't watched it before, you may find it helpful.

The thread i took that link from is here

Signal interference

What things/situations/etc. will interfere with the signal between my Mavic Pro and the controller? Thanks.

mavicpilots.com

It's unlikely that there was anything wrong with the calibration even though it was performed in the same location as the problematic takeoff. The magnetic distortion caused by reinforced concrete is typically very local - just within a few inches. Holding the aircraft while performing the calibration probably keeps it well out of the distorted field. Or, if the distortion is more extensive, then it's likely to be constant at the calibration location. Since the calibration is only subtracting out the magnetic field due to the aircraft, it actually shouldn't matter too much what the external constant field looks like.

 

[Starz]

I’m looking at the screenshot in post#3. Is the blue/green dot the launch point or is it the home point dot?
I’m hoping to educate myself a little on this one.
I read quite often about reinforced concrete sidewalks. In all the years I’ve seen sidewalks being built, I’ve not seen any using steel reinforcing in that process. I do understand that colder climate areas may be a different process in those respects.
So my question is, would it not be more of a vertical structure issue in this case? I would think there is much more steel above ground to provide the disturbance in this case.

 

[Deleted member 94047]

Sar, I do not pretend to know a tenth of the things you do. But the guy in the video I linked above has some very good arguments and it has convinced me enough not to trust the calibration would solve the problem if I am presented with a magnetic interference error. I would move to a different location before attempting a calibration.

 

[Meta4]

it has convinced me enough not to trust the calibration would solve the problem if I am presented with a magnetic interference error.

The name Compass Error confuses people into thinking they have to "fix" their compass.
A compass error would more correctly be called a compass warning.

The confusion continues with very few flyers understanding what calibrating the compass actually does.

 

[Meta4]

In all the years I’ve seen sidewalks being built, I’ve not seen any using steel reinforcing in that process. I do understand that colder climate areas may be a different process in those respects.

Here's an X-ray view of a sidewalk:

Steel reinforcing is standard for sidewalk construction in most places.
If it's constructed without reinforcing, it's going to crack horribly when subjected to any load or settling.

So my question is, would it not be more of a vertical structure issue in this case? I would think there is much more steel above ground to provide the disturbance in this case.

No .. the drone would have to be very close (maybe only a few feet) to the structural steel in the buildings.
But if the compass is just an inch or two away from a small steel item, that can do the damage.
Note in the image above, you might put the drone down in the centre of a square, or with the compass right above a crossing point, or where the mess panels are overlapped.
You don't know how much steel is within an inch or two of your compass.
Do you want to buy a ticket in that lucky dip?

 

[sar104]

Sar, I do not pretend to know a tenth of the things you do. But the guy in the video I linked above has some very good arguments and it has convinced me enough not to trust the calibration would solve the problem if I am presented with a magnetic interference error. I would move to a different location before attempting a calibration.

That's completely correct - I did not mean to imply that a calibration would solve the problem - it won't. I was simply pointing out that the calibration likely made no difference to the problem, which was a distorted magnetic field on the ground where the aircraft took off. The calibration was probably fine to begin with and fine after recalibration.