Crash, why/how did it happen?

[Djicase]

Hi all, I am quite experienced and had never any issues with my mavic 2 pro.
However. this night something really surprisingly happened: I starteted the drone in a backyard, GPS was not available in the first place so I took off in atti mode. No big deal as I just wanted to hover over the bbq 2-3 meters above ground. So I started the drone, turned left maybe 3 meters and just hovered. I did not touch anything at this moment as I just wanted to hover and leave the drone as it is. Then without any warning something happen I never experienced in 2 years of flying: the drone went kind of full throttle left and crashed into the wall of the backyard. No damage.

Does anyone have experienced something like this and can explain this?

 

[Djicase]

Ah well, after looking more into atti mode I think I simply got a victim of how atti mode functions. As I was geared on hovering and the distances in the backyard were too little the crash was just a logic consequence.

 

[Meta4]

Then without any warning something happen I never experienced in 2 years of flying: the drone went kind of full throttle left and crashed into the wall of the backyard. No damage.

Does anyone have experienced something like this and can explain this?

As you describe the incident it doesn't sound like what happens in atti mode at all.
In atti mode the drone has no horizontal position holding ability and can drift on any breeze and will continue drifting after you take your hands off the sticks (no brakes).
But it won't "zoom off at full throttle".
That suggests a different cause but the recorded flight data would be needed to tell.
Post your flight data and details of what your launch surface was and how close it was to any steel objects.

 

[Djicase]

Thx for your interest. Maybe you find out more than me.

 

[itsneedtokno]

Sounds like when you finally got GPS lock (ALWAYS WAIT) there was an issue with the compass.

What to take from this.

1) Wait for GPS lock
2) Check orientation in MiniMap
3) Record Home Point

 

[Djicase]

Indeed, that was my initial thought as well. That when it hovered a few meters above the backyard it received satellites/GPS and then it got "confused". Actually I was really lucky I just hovered in low enough altitude that it hit the wall. Because two meters higher and it would have passed the neighbours building and vanished into the urban surrounding with potentially extremely bad results.

 

[itsneedtokno]

Indeed, that was my initial thought as well. That when it hovered a few meters above the backyard it received satellites/GPS and then it got "confused". Actually I was really lucky I just hovered in low enough altitude that it hit the wall. Because two meters higher and it would have passed the neighbours building and vanished into the urban surrounding with potentially extremely bad results.

Yeah, my understanding (from reading TONS of posts on here) is that it will hover and whatnot just fine with a compass (or imu) that's askew, but once any input happens, or even a breeze makes the motors try to compensate the drone gets confused and launches itself in any random direction. The randomoscity (my own word lol) of the fly away can probably be determined, but I can't weigh in on that point.

 

[Meta4]

Thx for your interest. Maybe you find out more than me.

Here's a short summary of your flight data: DJI Flight Log Viewer - PhantomHelp.com

You had no GPS or a home point until 0:39.6

You made three small joystick inputs up to 0:14.1 and didn't touch the joysticks after that.
The speed was gentle and the drone came to a stop at 0:18.5.
At 0:38.5 (one second before the drone acquired GPS), the drone started moving without GPS input and the speed reached 1.7 metres/sec.
The collision occured at 0:41.6.

Because there was no GPS early on, we don't have full data for the early section of the flight.
But the drone moving without joystick input could have been caused by a yaw error which can happen when you power up the drone where magnetic distortion causes the drone's IMU initialisation to be wrong.

In post #3 I asked about the area where you powered up.
Can you describe it?

 

[boblui]

The same thing has happened to my Mini.

It was powered up close to steel reinforcement in the balcony floor where I launched. As the result the compass went WAYYYYYY off. It flew OK in the first minute or so because it couldn't get a GPS fix and it was flying in ATTI mode. As the compass is not used in this mode, the error didn't cause any problem. Then more satellites were seen and it finally went into P-GPS mode. Before long, it accelerated sideway all by itself and hit the balcony hard. Fortunately nobody was hurt. The movement was caused by the drone's attempt to correct it's position base on GPS. As the compass was pointing to the wrong direction, the more it tried to correct, the bigger the position error became so the FC got anxious and commanded the drone to fly faster and faster before collision occurred. As you can see from the video below, it banked so hard that the gimbal reached it's limit of travel and could not maintain the camera horizontal.

If you can provide the .DAT log file, the mystery can be resolved. I bet there is a big discrepancy between the IMU yaw and magnetic yaw

 

[Djicase]

The area is an urban residential area. Usually electric cables are always laid below ground. If you look at the log DJI Flight Log Viewer - PhantomHelp.com you can see the maps picture of the backyard with the trajactory of the drone.

However, I cannot 100% rule out a magnetic field of course, though I don't know where it could come from. But it's not my backyard, so maybe there is a source I don't know about.

 

[Djicase]

The same thing has happened to my Mini.

It was powered up close to steel reinforcement in the balcony floor where I launched. As the result the compass went WAYYYYYY off. It flew OK in the first minute or so because it couldn't get a GPS fix and it was flying in ATTI mode. As the compass is not used in this mode, the error didn't cause any problem. Then more satellites are seen and it finally went into P-GPS mode. Before long, it accelerated sideway all by itself and hit the balcony hard. Fortunately nobody was hurt. The movement was caused by the drone's attempt to correct it's position base on GPS. As the compass was pointing to the wrong direction, the more it tried to correct, the bigger the position error became so the FC got anxious and commanded the drone to fly faster and faster before collision occurred. As you can see from the video below, it banked so hard that the gimbal reached it's limit of travel and could not maintain the camera horizontal.

If you can provide the .DAT log file, the mystery can be resolved. I bet there is a big discrepancy between the IMU yaw and magnetic yaw

wow, that is extremely interesting. From the course of events it looks really 100% like what I had yesterday. So I will have a look at the backyard next time again and scan for some huge metal areas. Although I am really surprised where that could be located. It's a typical Brussels residential area and houses are all made of bricks. But of course there might be something like roof/balcony/terrace construction that caused the crash.

 

[Meta4]

wow, that is extremely interesting. From the course of events it looks really 100% like what I had yesterday. So I will have a look at the backyard next time again and scan for some huge metal areas. Although I am really surprised where that could be located. It's a typical Brussels residential area and houses are all made of bricks. But of course there might be something like roof/balcony/terrace construction that caused the crash.

What was the launch surface?
Reinforced concrete contains lots of steel and launching from reinforced concrete is the most common cause of what you saw.
Most end even worse.

 

[BadWolf1]

You had no GPS or a home point until 0:39.6

What I find odd, is the ODS:IsGPSUsed flag goes to true at 9.4 seconds but general:lat/lon are 0 until 38.3 and homepoint is not set until 39.6

The drone seems make a move without any control inputs between these two. Did acquiring a lat/lon fix cause that ?

 

[Meta4]

What I find odd, is the ODS:IsGPSUsed flag goes to true at 9.4 seconds but general:lat/lon are 0 until 38.3 and homepoint is not set until 39.6

The drone seems make a move without any control inputs between these two. Did acquiring a lat/lon fix cause that ?

The drone was assessing the quality of the GPS data and rated it as zero until just before the home point was recorded.
Normally getting GPS would only leave the drone hovering in place.
It wouldn't cause any issue.
But if the drone was initialised incorrectly in a distorted magnetic field, the drone can move off on its own just trying to hold position.
Because of the incorrect initialisation, the direction it moves is wrong and it gets further way, corrects again and again and again, getting further away and faster as it goes.

 

[Deleted member 114292]

Hi all, I am quite experienced and had never any issues with my mavic 2 pro.
However. this night something really surprisingly happened: I starteted the drone in a backyard, GPS was not available in the first place so I took off in atti mode. No big deal as I just wanted to hover over the bbq 2-3 meters above ground. So I started the drone, turned left maybe 3 meters and just hovered. I did not touch anything at this moment as I just wanted to hover and leave the drone as it is. Then without any warning something happen I never experienced in 2 years of flying: the drone went kind of full throttle left and crashed into the wall of the backyard. No damage.

Does anyone have experienced something like this and can explain this?

If you’re gonna fly and Atti mode you should do the MOD so you have full-time A mode.
Because if you takeoff without GPS weird things happen as it starts to go in.
When I fly in A mode I always get GPS lock first so if something goes awry I can switch back.

Also you should never fly Atti mode in a confined space.

 

[Djicase]

What was the launch surface?
Reinforced concrete contains lots of steel and launching from reinforced concrete is the most common cause of what you saw.
Most end even worse.

Launch area was a terrace with wooden tiles. However, it is indeed very possible that the wood tiles rest on reinforced concrete. I remember that I had no sufficient signal and I went on the loan. There it got a signal. Due to the hight of the gras I didnt to launch there and went back to the terrace. There it went back to atti mode and I launched anyway.

 

[Lapeer20m]

If you’re gonna fly and Atti mode you should do the MOD so you have full-time A mode.
Because if you takeoff without GPS weird things happen as it starts to go in.
When I fly in A mode I always get GPS lock first so if something goes awry I can switch back.

Also you should never fly Atti mode in a confined space.

this!

I have flown original mp from outside to inside many times (with prop cages) and there is all kinds of uneasiness when the aircraft switches from gps to vision, sometimes not accepting any stick inputs for severalSeconds while it just does what it wants to do.

a bit unsettling the first time.

i think the same can also h when it switches back from
Vision to gps

i know you mentioned atti mode, but it was likely in vision mode.

 

[Djicase]

In the log you can see that I started in atti mode. Then it switched to p-gps. In second 38-39 you can see that the drone suddenly shifts to the right (from the drone point of view) without any stick command made.

Thx for all the comments. For now my lessons learnt are:

- avoid flying when there is too weak gps signal
- if signal is weak and you still want to fly switch on a-mode permanently in order not to risk that gps is kicking in later as this could lead to trouble.
- atti mode demands sufficient buffer zones for manoeuvring when using it outside. A breeze can drift the drone quickly.
- if a magnetic field is expected, e.g. from large metallic areas in close distance, do not fly or switch on atti mode to bypass the compass and gps.

 

[Deleted member 114292]

In the log you can see that I started in atti mode. Then it switched to p-gps. In second 38-39 you can see that the drone suddenly shifts to the right (from the drone point of view) without any stick command made.

Thx for all the comments. For now my lessons learnt are:

- avoid flying when there is too weak gps signal
- if signal is weak and you still want to fly switch on a-mode permanently in order not to risk that gps is kicking in later as this could lead to trouble.
- atti mode demands sufficient buffer zones for manoeuvring when using it outside. A breeze can drift the drone quickly.
- if a magnetic field is expected, e.g. from large metallic areas in close distance, do not fly or switch on atti mode to bypass the compass and gps.

That sounds right. Some where there is a tutorial from sar about how to add ATTI mode.
You should go to like 80 feet to practice or whatever altitude is safe where you are flying,

 

[sar104]

That sounds right. Some where there is a tutorial from sar about how to add ATTI mode.
You should go to like 80 feet to practice or whatever altitude is safe where you are flying,

Mavic 2 Atti mode

I haven't seen anyone else mention it so, just in case anyone is interested, I can confirm that it is possible to switch the M2P into ATTI mode by changing the mapping on the flight mode switch using the same DJI Assistant functionality as with the MP. I altered the Sport mode setting (8) to...

mavicpilots.com