Out of control drifting and crash into wall

[saigonsaigon]

Hi everyone, first post on the forum (and first crash) after 3 years with DJI drones. My 2 months old Mavic Pro 2 decided to do some drifting today and crashed in a wall.

I was shooting between two factory buildings at a low height (corporate shoot) and instead of hovering like normal, the drone would drift heavily from side to side.
Was in tripod mode to take a shot and the drone decided to drift quickly to the right and crashed into the factory building, thankfully at a very low height (about 2 meters), Did a flip and crashed on its top, battery flying out, top cover will need to be replaced but the camera and gimbal seem fine for the moment, which is a relief.
No specific warnings about compass or magnetic interference but I figure it could have been that.

The side sensors were completely useless, didn't even notice the wall - the drone just drifted 4 or 5 meters straight into the wall, before I had a chance to go full throttle to the left.

On the flight log you can see the flight ending abruptly, without landing. I guess the "upwards obstacle detected" is the ground since the drone had flipped right before crashing.

First real problem with a DJI drone but quite a worrying one, the crash could have done much more damage to the drone.

Flight log : DJI Flight Log Viewer - PhantomHelp.com

 

[calmar81]

You weren't indoors but maybe a bad satellite cover was the problem for your crash. I had an indoor crash two weeks ago (thread: First crash – indoors) and as well the users as DJI in my repair quotation mentioned, that the drones heavily (or solely) rely on GPS data to maintain position.

Maybe there was a magnetic intereference between those two buldings but I think your and my crash may have had the same cause.

 

[saigonsaigon]

Thanks for your answer. I've flown indoor many times, with no satellite, only Vision. Neither my old Phantom 4 or the Mavic 2 has ever done something like this before. This wasn't a slight drift by 1 or 2 meters but at least 4 or 5 without stopping (and in tripod mode). With 15 satellites active at time of crash according to the flight log.
Seems very strange. On the same corporate shoot I did plenty of shots inside warehouses and it was perfect, not drifting one bit.

But maybe it's something unpredictable ! And like you say on your post, it is quite a resistant drone. I was surprised the damage wasn't worse.

 

[NorthGADronePilot]

Regardless, since he was in tripod mode, shouldn't one of the many omnidirectional sensors have "noticed" the wall and halted the bird before it unceremoniously crashed?

 

[saigonsaigon]

That's the biggest question for future shoots. It was a huge concrete wall, obstacles don't get bigger than that. And sensors did absolutely nothing. It just went straight for the wall.

 

[FoxhallGH]

At the time of the crash, it looks like 15 sat's were in view for GPS, but the drone had detected an overhead obstruction just before ...
[Edit - sorry just noted your original comment about the flipping of the drone causing the 'overhead obstruction' !!!]

 

[saigonsaigon]

Yes, the overhead obstruction is because the drone flipped after hitting the wall. So the "overhead" was the ground. Basically 1 second before crashing into it.

 

[HoozierDroneDaddy]

If you were between two large factory buildings, were they active? Did you check to see if they have an active WiFi or other radio system in the area of your flight that could interfere with your system? When flying car lots, that is one of the pre-checks, and you have them shut it down during the flight.

 

[saigonsaigon]

I always leave them on. Can the sensors switch off automatically ?
Definitely nothing like WIFI or radio system that could have interfered with the system. I flew for 2 days in the area, connection was perfect all the time, not a problem. Except on that flight.
I've never had a problem where the drone just moves in a direction by itself, with no input.

 

[NorthGADronePilot]

This is extremely disturbing to me. I don't care if you're locked on 15 satellites or 0, those sensors should keep you from smashing into a building wall. If it were a tree branch, power line, or some other thin obstruction, I could understand the failure to detect.

 

[saigonsaigon]

Screenshot from a shot right after, the drone crashed against that wall, in the red circle. Sensors were for sure activated, unless tripod mode automatically deactivates them, but they were working perfectly fine 30 minutes before.
If the sensors can't sense a wall like that, they probably can't sense much !

 

[Mavic Ray]

The sensors are basically cameras so they are visual. I wonder maybe if they couldn't detect that wall because maybe that wall had no features or and depth/contrast. So to the sensors/cameras the wall just looked like open sky?

 

[Simmo]

I would bet magnetically distorted environment for take off. Thats a lot of concrete around you there???
@sar104

 

[calmar81]

I would bet magnetically distorted environment for take off. Thats a lot of concrete around you there???
@sar104

I think, also the street is made out of reinforced concrete with plenty metal in it for heavy machines / trucks etc. Maybe also power-lines and gas-pipes in there?

But then a compass error should have shown up prior to take up... and @saigonsaigon said, he flew multiple times on that particular location without an error.

 

[saigonsaigon]

I would bet magnetically distorted environment for take off. Thats a lot of concrete around you there???
@sar104

I think that is probably the most likely explanation. But no mention on app, and the sensors didn't detect the wall. That's a lot of combined problems for what should have been a simple shot.

Re calmar81, no power lines and the staff said there shouldn't be much metal but could have been.
And yes all other flights in the area we're perfectly fine, even those completely inside the warehouses.

I've sent everything to DJI, we'll see what they say.

 

[calmar81]

I've sent everything to DJI, we'll see what they say.

Good luck with that!

 

[sar104]

I would bet magnetically distorted environment for take off. Thats a lot of concrete around you there???
@sar104

I'd need to see the mobile device DAT file to be certain, but the GPS location and the IMU data do not agree at all. Normally, with reasonable GPS satellite numbers and reported GPS health I would suspect magnetic interference at the takeoff point, but there are some strange dropouts in the GPS data. It could be either.

Here is the comparison of GPS and IMU-derived position data, showing the problem:



I recommend posting the DAT file:

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

 

[SteveCA]

I had a very similar thing happen. while landing last weekend. I was manually landing my MP2 and just after it touched down, instead of the motor shutting down (left stick was pulled all the way back as usual) the motor kept running and the drone lifted up about 5 inches and drifted quickly to the right. I caught it and stopped the drift, put it back up and hand caught it. When I put it back on the concrete floor, It reported compass interference.

I’ve flown from that exact spot at least a dozen times with no issue but that day, it showed compass issues. What I can’t understand is why the motor sped up while the left stick was all the way down after landing.

This is the first malfunction I’ve experienced with any DJI drone in over 3 years and over 700 flights.

 

[sar104]

I had a very similar thing happen. while landing last weekend. I was manually landing my MP2 and just after it touched down, instead of the motor shutting down (left stick was pulled all the way back as usual) the motor kept running and the drone lifted up about 5 inches and drifted quickly to the right. I caught it and stopped the drift, put it back up and hand caught it. When I put it back on the concrete floor, It reported compass interference.

I’ve flown from that exact spot at least a dozen times with no issue but that day, it showed compass issues. What I can’t understand is why the motor sped up while the left stick was all the way down after landing.

This is the first malfunction I’ve experienced with any DJI drone in over 3 years and over 700 flights.

The flight log is your friend.

 

[Colbymh]

So ur saying it jumped out of tripod mode also besides the sensors not working? Cuz there is nothing quick about tripod mode. Lol