Runaway Mavic 3 Cine

[Deleted member 162835]

Well a sad event happened today. We took the Mavic 3 out down by the Warhol Bridge and the drone went nuts. When it took off a straight command resulted in the drone shooting up and out above the Ohio river (you know this is going to be bad). Subsequent commands to fly the drone straight resulted in it further climbing. I decided to bring it back because of the erratic behavior. Commands to fly it back resulted in the drone dropping in altitude. I paused for the holy $hit moment and tried again to command it forward only to watch it dive bomb into the river full speed. The drone was in N mode with nothing around. Wind conditions were nominal and not a factor. I've flow the drone numerous times before this even in really rough wind conditions without much of an issue. Today was an exception. I know water reflections could upset the sensors but would it do so to the point of the drone ignore commands and doing opposite user input? Drone was fully updated BTW.

Is it possible to pull flight logs of the RC Pro? I've never tried that. I do have care refresh and filed a case with DJI which is being investigated. Not in a happy mood after this one.

 

[vindibona1]

I'm sorry. While I'm sure this isn't a regular occurrence with others, one has to take pause before spending $5k on a drone that could be destroyed or lost in an instant. OTOH, if you are a hard core professional needed the best footage you can get within a semi-affordable price, I suppose you have to take a chance.

I think when you get up into that price range hull/fly away insurance is almost mandatory.

 

[Deleted member 162835]

I'm sorry. While I'm sure this isn't a regular occurrence with others, one has to take pause before spending $5k on a drone that could be destroyed or lost in an instant. OTOH, if you are a hard core professional needed the best footage you can get within a semi-affordable price, I suppose you have to take a chance.

I think when you get up into that price range hull/fly away insurance is almost mandatory.

Yes. The drone is insured so I'm not concerned with getting another one. All of my photography gear is covered for the very reason to cover unexpected loss. It's the erratic controls and lack of response to user input that concerns me. The drone has never acted like that. Those around agreed it was out of control. I brought this drone to use it and will use it. The replacement will be used as well with a little more caution.

 

[Deleted member 162835]

I just pulled the logs which show user input into the controls. When I was bringing the drone back I kept telling it to go up and the flight recorder shows that but the drone dived bombed into the water. The more I think on this I'm glad it crashed into the water. It could have lost control into another person that was around and hurt someone. I think DJI will be giving me a new drone and battery but this is not a good event to have occured. We shall see.

 

[Meta4]

I just pulled the logs which show user input into the controls. When I was bringing the drone back I kept telling it to go up and the flight recorder shows that but the drone dived bombed into the water. The more I think on this I'm glad it crashed into the water. It could have lost control into another person that was around and hurt someone. I think DJI will be giving me a new drone and battery but this is not a good event to have occured. We shall see.

Can you post the recorded flight data please?
You can get some feedback and find out if there was something that caused the issue or if it was a genuine fault of teh drone.

 

[GDadd]

Well a sad event happened today. We took the Mavic 3 out down by the Warhol Bridge and the drone went nuts. When it took off a straight command resulted in the drone shooting up and out above the Ohio river (you know this is going to be bad). Subsequent commands to fly the drone straight resulted in it further climbing. I decided to bring it back because of the erratic behavior. Commands to fly it back resulted in the drone dropping in altitude. I paused for the holy $hit moment and tried again to command it forward only to watch it dive bomb into the river full speed. The drone was in N mode with nothing around. Wind conditions were nominal and not a factor. I've flow the drone numerous times before this even in really rough wind conditions without much of an issue. Today was an exception. I know water reflections could upset the sensors but would it do so to the point of the drone ignore commands and doing opposite user input? Drone was fully updated BTW.

Is it possible to pull flight logs of the RC Pro? I've never tried that. I do have care refresh and filed a case with DJI which is being investigated. Not in a happy mood after this one.

Do you know whether you acquired enough satellites before flying? A friend took off with his Mavic 3 Cine before getting to the minimum and immediately lost control and crashed. He used his refresh to receive a replacement.

 

[DougMcC]

Another key preflight in my M2P is confirmation of compass heading. I never take off until the acquired satellites indicator is greater than 12 or so, the compass arrow is properly aligned on the map AND that Arrow moves appropriately to the left and right as I turn the aircraft left to right. Finally, after motor start or earlier I must hear the drone mistress say the magic words about the Home Point. THEN I will fly. That initial compass heading check can be the key to hearing “have a nice day” Well she doesn’t actually say that. By posting your flight files you can tell about the satellite locks and compass issues. Sorry about loosing your bird.
Being certain about satellite lock and a functioning correct initial compass heading are key to every successful flight.

 

[Meta4]

Do you know whether you acquired enough satellites before flying? A friend took off with his Mavic 3 Cine before getting to the minimum and immediately lost control and crashed.

Being impatient and launching early doesn't cause the drone to crash.
Even with no GPS it should be fully controllable.
Without seeing the data we won't know what the problem was, but it wasn't lack of GPS.

 

[DougMcC]

Being impatient and launching early doesn't cause the drone to crash.
Even with no GPS it should be fully controllable.
Without seeing the data we won't know what the problem was, but it wasn't lack of GPS.

But a misaligned compass at launch can rapidly create flight chaos can it not? And yes we need the data to know

 

[Meta4]

But a misaligned compass at launch can rapidly create flight chaos can it not? And yes we need the data to know

A yaw error is completely different from not having GPS.
Not having GPS does not cause a loss of control.

 

[Starz]

Whenever I see the word”bridge”and a DJI product together, I cringe.

 

[GDadd]

Being impatient and launching early doesn't cause the drone to crash.
Even with no GPS it should be fully controllable.
Without seeing the data we won't know what the problem was, but it wasn't lack of GPS.

In the case of the person I mentioned, he wasn't being impatient. He was new to the platform and didn't know to wait.

I agree with DougMcC and follow the same ritual. Getting GPS was shorthand for waiting for the M3 to be controllable as expected and not meant to be misleading.

 

[bravezoo63]

Had the stick mode on the Fly app control tab been inadvertently changed? Just a thought. So sorry for your loss.

 

[Starz]

A friend took off with his Mavic 3 Cine before getting to the minimum and immediately lost control and crashed.

Very unlikely this incident had anything to do with the number of satellites the friend did or didn’t have.

 

[Deleted member 162835]

Can you post the recorded flight data please?
You can get some feedback and find out if there was something that caused the issue or if it was a genuine fault of teh drone.

Flight data attached. I found a better analysis program than reviewing on my RC Pro that broke the inputs down into finer granularity. At the end I did do an input down for a brief moment but that was after it was gone. At 42.6 I was trying to send it back up and it was not going. The flight was brief because and I tried to bring it back because of the erratic behavior. Note, I have the right stick controlling up and down movement (thrust) along with yaw. The left is controlling roll and pitch. I don't think the drone new where it was at in relation to space and again the controls resulted in erratic behavior.

After a day of lamenting on it I've come to peace with the event. I'll pay to use my fly away coverage if DJI determines user error and will be WAY more cautious in the future.

 

[Deleted member 162835]

Do you know whether you acquired enough satellites before flying? A friend took off with his Mavic 3 Cine before getting to the minimum and immediately lost control and crashed. He used his refresh to receive a replacement.

Yeah satellite lock was fine. I won't launch unless I have sufficient lock. It did for a moment dip down during the flight to like 9 or 10 satellites but went back up quickly.

 

[Meta4]

I agree with DougMcC and follow the same ritual. Getting GPS was shorthand for waiting for the M3 to be controllable as expected and not meant to be misleading.

Having no GPS does not make a drone uncontrollable.
With no GPS, it's still fully controllable, but might not have horizontal position holding..

 

[Meta4]

Yeah satellite lock was fine. I won't launch unless I have sufficient lock. It did for a moment dip down during the flight to like 9 or 10 satellites but went back up quickly.

GPS was not fine at the start of your flight.
Your GPS reliability was low (2/5) until 17 seconds into the flight.
But this did not cause the problem.

Flight data attached.

Here's a summary and visual representation of the flight:

DJI Flight Log Viewer | Phantom Help

Upload and view flight logs from your DJI drones

www.phantomhelp.com

The first indications of a problem show up in the compas data 4.4 seconds into the flight.
Following this, the drone doesn't appear to properly respond to rudder input and stays on a heading of 87° despite joystick input that should have rotated the drone anti-clockwise first, and then clockwise.

At 8.8 seconds the drone starts to acellerate without any right stick input.
It is erratic and slows and speeds up on its own.

At 20 seconds there is an umcommanded descent from 47 ft to 22 ft.
The cause of this is quite puzzling.
There is another small descent at 37.4 sec when the joystick input should have caused the drone to climb.

Can you remember which way the drone was pointing while it was on the ground waiting for launch?
The data shows it to be just east of north (27°) but I suspect it was something else.
It looks like you launched from a reinforced concrete surface (never a good idea) and this might explain part of the flight problems.

@sar104 might be able to help with further analysis

 

[sar104]

GPS was not fine at the start of your flight.
Your GPS reliability was low (2/5) until 17 seconds into the flight.
But this did not cause the problem.


Here's a summary and visual representation of the flight:

DJI Flight Log Viewer | Phantom Help

Upload and view flight logs from your DJI drones

www.phantomhelp.com

The first indications of a problem show up in the compas data 4.4 seconds into the flight.
Following this, the drone doesn't appear to properly respond to rudder input and stays on a heading of 87° despite joystick input that should have rotated the drone anti-clockwise first, and then clockwise.

At 8.8 seconds the drone starts to acellerate without any right stick input.
It is erratic and slows and speeds up on its own.

At 20 seconds there is an umcommanded descent from 47 ft to 22 ft.
The cause of this is quite puzzling.
There is another small descent at 37.4 sec when the joystick input should have caused the drone to climb.

Can you remember which way the drone was pointing while it was on the ground waiting for launch?
The data shows it to be just east of north (27°) but I suspect it was something else.
It looks like you launched from a reinforced concrete surface (never a good idea) and this might explain part of the flight problems.

@sar104 might be able to help with further analysis

So far I can't figure out what was going on. There are significant inertial errors that are clear in both the velocity and position data:



That could be the result of a compass error on takeoff, but it doesn't look quite right. As noted by @Meta4, the indicated heading at takeoff was 27°, and it would be very useful to know if that is correct.



The stick data lines up poorly with the aircraft response. It's clear that the left stick (up-down), which is throttle in default mode, was mostly controlling pitch:



The left stick (left-right), default rudder, was controlling roll, at least some of the time:



The right stick (left-right) was controlling yaw except during the first 20 s of the flight when the FC appears to have been commanding CW rotation - possibly in an attempt to resolve a yaw error:



But the remaining axis, the right stick (up-down), by default the elevator, is not controlling the vertical motion:



It doesn't look like a typical z-axis accelerometer problem, since the height lines up at least moderately well with the negative z axis velocity integrated with respect to time.

It's a very unscientific conclusion, but the FC looks really sick on that flight.

 

[kodiak]

<<SNIP>>
It looks like you launched from a reinforced concrete surface (never a good idea) and this might explain part of the flight problems.

As someone who flies in urban areas regularly with a variety of drones (from DJI and others) this comment is interesting. I understand how rebar or other ferrous metal could influence and corrupt the compass but are you saying it can also cause other kinds of harm?