Mavic Air 2 - Unresponsive and crash

[xanthosc]

Hello people. I just crashed my MA2 and I am in search for some answers. While doing a pull back, the drone got unresponsive and continued flying backwards, took a slight curved path, till it hit an obstacle (luckily was not a head!). I want to figure out what happened since this is not the first time it happened. I had a similar incident some weeks back but during that flight I was flying with a beta 3rd party software and an older phone and so I assumed that this was the cause. Can anybody interpret the log files and confirm if the drone was unresponsive or if the crash was due to any improper controls from my side. Today I was using the drone's native app and a Samsung Note 8.

20 minutes before the accident, I had successfully flew 2 other flights.

From a safety point of view, it is very critical to know what's going on since accidents like these can result in unspeakable consequences.

The video of the incident can be found below

Dropbox - File Deleted - Simplify your life

www.dropbox.com

Thank you
Xanthos

 

[Phantomrain.org]

Magnetic interference and weak gps signal comes to mind, you can almost see the exact moment in the tunnel when the compass got hammered. Those are tough flying conditions for anyone and not much you can do better but be prepared for Atti mode.

Phantomrain.org
Gear to fly your Mavic in the Rain / Snow

 

[slup]

Hello people. I just crashed my MA2 and I am in search for some answers. While doing a pull back, the drone got unresponsive and continued flying backwards, took a slight curved path, till it hit an obstacle (luckily was not a head!). I want to figure out what happened since this is not the first time it happened. I had a similar incident some weeks back but during that flight I was flying with a beta 3rd party software and an older phone and so I assumed that this was the cause. Can anybody interpret the log files and confirm if the drone was unresponsive or if the crash was due to any improper controls from my side. Today I was using the drone's native app and a Samsung Note 8.

20 minutes before the accident, I had successfully flew 2 other flights.

From a safety point of view, it is very critical to know what's going on since accidents like these can result in unspeakable consequences.

The video of the incident can be found below

Dropbox - File Deleted - Simplify your life

www.dropbox.com

Thank you
Xanthos

Nope ... it wasn't magnetic interference that caused this.

This is not a mystery at all ... & it was a clear pilot error. You flew into that chimney which covered most of the sky which meant that you got a very bad GPS coverage ... this together with the low light made your VPS sensors not working properly... you was probably just seconds away from ATTI mode.

 

[xanthosc]

Being unable to use the sensors due to light conditions, I used tripod to slow down the drone's reactions and smooth out my commands but I was not aware that without gps coverage, the controller and the drone at 15m apart without any barrier between them it would hinder command transfer. Even when it exited the vessel, I was pushing it to gain altitude without any response.
All these happened in seconds.
Is this an expected behavior when there is no gps coverage to have totally unresponsive sticks? If yes, I learnt it the hard way!!

 

[xanthosc]

Also, from AirData I get this

Mavic Air 2/iOS
DJI Fly 1.2.2

Any thoughts why it says iOS when in reality I am using android?

 

[slup]

Being unable to use the sensors due to light conditions, I used tripod to slow down the drone's reactions and smooth out my commands but I was not aware that without gps coverage, the controller and the drone at 15m apart without any barrier between them it would hinder command transfer. Even when it exited the vessel, I was pushing it to gain altitude without any response.
All these happened in seconds.
Is this an expected behavior when there is no gps coverage to have totally unresponsive sticks? If yes, I learnt it the hard way!!

It's not unusual to have a mushy stick response & a erratic flight behavior in situations when the positional confidence from the GPS is really low & at the same time not having horizontal stabilization assistance from the VPS sensors. That's why you got those warnings on the app screen.

Wonder if you planned for the risk of loosing all horizontal positional hold ... no satellites & no VPS lock when you entered that chimney or if you know what ATTI mode is & how to master it?

 

[eaglesurfing]

Wow.
But drones can go anywhere right?

 

[xanthosc]

No reason to get critical on what I know or don't know, planned or not planned.
I am aware of ATTI mode but I am obviously an inexperience pilot. This is exactly why I am communicating here. To find out what went wrong and make sure that I will not repeat the same mistake. Also, I am confident that other inexperienced pilots will learn something from this and maybe avoid a similar mistake.
Having said that, I have carried out similar flights in confined spaces in the past without any issues. The main difference was that in previous flights I was in concrete surrounding and not metal like today. So maybe Phantomrain has a really valid point above.
Thank you for your insides on the matter. Much appreciated.

 

[slup]

No reason to get critical on what I know or don't know, planned or not planned.
I am aware of ATTI mode but I am obviously an inexperience pilot.

No criticism intended there ... I'm just wondering because I suspected that you were a inexperienced pilot.

Trying to pull off a flight like that requires that you plan for & expect drifts & ATTI mode and that you can master those conditions like what you encountered there.

...I have carried out similar flights in confined spaces in the past without any issues.

Can't say anything about those other flights without logs ... but if you pulled them off you either had a better GPS reception than you thought or you was closer to ground with better light so the VPS sensors could stabilize the craft horizontally. Do try to upload the logs from those flights to Airdata.com yourself ... it's free, & check how they look. For the Mini 2 & with a FLY app from v1.2.2 Airdata is for time being the only that exists to check off flight logs.

The main difference was that in previous flights I was in concrete surrounding and not metal like today. So maybe Phantomrain has a really valid point above.
Thank you for your insides on the matter. Much appreciated.

No, this wasn't magnetic interference ... If your craft's compass was magnetic interfered at power on & by that initialized the IMU wrongly you shouldn't have been able to get out of that chimney with the Mini 2 in one piece.

If the area you passed on that flight was magnetic interfered on the other hand, that hadn't caused any problems during the short time you were in the chimney ... the chimney may have disturbed the compass yes, but it's not that, that's stabilizing the craft ... it's the IMU, so if that was correct after power on, a compass starting to be deflected during flight should only slowly together with the gyro feed in corrections to the IMU. The time was way to short for that to happen.

No ... this was only due to a craft that was nearly blindfolded position wise ... that affected the craft's positional performance & ability to notice drifts & measure the results from your stick inputs.

 

[boblui]

I don't think the crash was due to weak GPS signal or the pilot unable to control the craft in quasi-ATTI mode.

Near the end of the flight, the caft was already outside the tube. The GPS signal was not great but vision positioning should have kicked in. Then suddenly the direction of the craft changed by about 150 degrees without any rudder input :



As seen in the video, the direction of the craft did not change much in reality ( still pointing towards the entrance of the tube ) so the yaw change perceived by the flight controller was clearly due to sensor error. With such a large yaw error, fly away is ensured and your case is no exception . The log as well as the video indicated that the craft subsequently banked to the right by itself without any aileron input before crashing.



Incidents of this kind have been reported in the past. For some unknown reasons, the IMU suddenly made a wrong adjustment to the yaw reading mid flight causing the craft to fly away in an uncontrollable manner.

If the craft is still under warranty, you may consider filing a claim to DJI.

 

[slup]

Completely without access to any DAT log or even a proper method to interpret the TXT log I stay away from pointing this to hardware errors in this stage. What's noted though is a flight environment all setup for a very sketchy positional hold which the log confirms both regarding the sat count & that the lightning was weak ... then the log shows one more thing that also is usual when a position is missing.

Here at 17m 55,4sec & 73,3m from HP



Here 0,3sec later ... now 95,9m from the HP.

Note the large jump regarding the AC position, that transportation equals to over 75m/s.



The clip doesn't seem to show the crash itself ... suspect that the turns seen in the log in the end originate from the crash & not a sensor error.

Only DJI can say for sure as they have access to all information.

 

[Meta4]

While doing a pull back, the drone got unresponsive and continued flying backwards, took a slight curved path, till it hit an obstacle

From a safety point of view, it is very critical to know what's going on since accidents like these can result in unspeakable consequences.

This is exactly why I am communicating here. To find out what went wrong and make sure that I will not repeat the same mistake.

Your issue was that you flew in an environment where you completely lost the benefits of horizontal position holding.
Without that, the drone will continue drifting when you centre the sticks, like driving on ice.
It wasn't unresponsive, just didn't respond in the way you expected, because it had no "brakes".

From 1045-1076 seconds and again from 1083 seconds to the end of data, your data shows GPS Health of <4.
When GPS Health is <4 the flight controller has no confidence in the accuracy of teh GPS information and does not use it.

Having said that, I have carried out similar flights in confined spaces in the past without any issues. The main difference was that in previous flights I was in concrete surrounding and not metal like today.

Without seeing what the GPS Health was for those flights, you can't say that they were similar.
ps .. concrete structures are half steel

So maybe Phantomrain has a really valid point above.

That would be highly unlikely.
All he did was make a guess with no knowledge of what the data shows.

 

[unionglzr]

No reason to get critical on what I know or don't know, planned or not planned.
I am aware of ATTI mode but I am obviously an inexperience pilot. This is exactly why I am communicating here. To find out what went wrong and make sure that I will not repeat the same mistake. Also, I am confident that other inexperienced pilots will learn something from this and maybe avoid a similar mistake.
Having said that, I have carried out similar flights in confined spaces in the past without any issues. The main difference was that in previous flights I was in concrete surrounding and not metal like today. So maybe Phantomrain has a really valid point above.
Thank you for your insides on the matter. Much appreciated.

Don't kid yourself concrete has steel rebar in it which can interfere .....and don't be so sensitive

 

[Meta4]

But anyway, have you tried calibrating your IMU, Remote Controller and drone in a location AWAY from metal and magnets? What about restarting the phone and app? This should do the trick, but if not, the app may be a problem.

You've guessed everything that is unrelated to the actual issue.
None of that would have made any difference.
But the data clearly showed what the cause of his problem really was - see post #12 for an explanation.

 

[Meta4]

Maybe that could be an issue too, but I know the things I suggested would work. Ive tried it.

It's very common for forum members to throw out guesses like that without any knowledge of what's in the data.
What you've suggested wouldn't have been any help since none of it relates to what the data showed the actual problem to be.
Rather than tossing guesses around, leave it to the people that actually read the flight data and know what they are talking about.

That's how things work with airplane accident investigations.
The same applies to drone incidents.

 

[Meta4]

I am not just throwing guesses out like that. I have prior knowledge with these issues.

The flight data clearly shows the actual cause of his incident.
If you'd looked at his flight data, you would have commented on what caused the incident and how your "suggestions" might help.

But you didn't look at the data and don't know what caused the incident so just guessed the standard things that are suggested by people that don't understand these drones or how they work.
None of what you suggested would make any difference to the actual issue and the early parts of his flight confirm that none of those things were causing any problem.

 

[Mike Mas]

Weak or loss of GPS positioning does not render the drone to be un-controllable nor cause it to strike an object. It simply means the pilot now has to fly the equipment. Even if the model was flown back to GPS coverage, the process to re-acquire and stabilize the model is not an instantaneous process.

Mike