Hi everyone!

Finally happened to me, what I have been always affraid - got an uncontrollable flyaway with my Mavic Air.
Yesterday, at around 15:00 CET time, I took off for a forest filming, from a walley. Got 11 satellites, good GPS signal. Slowly climb up in the walley. At a few seconds after takeoff I got a Backward Vision sensor calibration error, which I ignored (bad decision!! ) thinking, ehh never needed that, I don't trust obstacle avoidance anyway.  Nothing suspicious, started recording.

When I reached about 250m, suddenly the image started to shake. I stopped the drone and watched the screen for warnings, nothing on the screen, I decided to turn back but when in looked on the screen saw the horizon tilted, and the drone flying sideways. I tried to locate the dron by visual, and figure out it's orientation. I saw it flying at tremendous speed to the other side of the walley, tried to climb up to gain attitude and to avoid the hill on the others side, but it was too late and crashed to the hillside, luckily in some bushes, ended up with two broken propellers, nothing else harmed. When I get back to home, I analized the Flight record, in I saw that the dron started to fly sideways andd accelerated to up to 130 (!!!) kph.. (I think I own the land speed record for Mavic Air from now).

After one hour of search in managed to recover it from the bushes, unharmed except 2 propellers. I suffered a lot of scratches and muscle fewer this morning, when climbig up my fat a** to a 100m high steep hill thorough the thick spiny bushes...
On on side I'm happy to recover my drone, but on the other hand I lost confidence in flying it, wht if this phenomenon will happen again.
Other details that should be mentioned:

Cold weather, -7 degress, low humidity. Drone and battery was in the car, at 20 - 22 degrees so no temp issues. I flown exactly in this place earlier this morning, on exactly the same path, with absolutelly no issues.
I suspect an accelerometer or gyro issue (which explains the camera shake right before the flyaway) but not sure about is.  Too bad the recording stopped when the flyaway begun, don't remember if it was stopped by me or by the system.

Phantomhelp log viewer here:
https://www.phantomhelp.com/logviewer/FJYVCD6GAVRKA2Q7WSFZ/

Can somebody explain what was the reason? I'm affraid to fly this drone until I clearly know exactly what happened. Who knows maybe next time I won't be so lucky and crash it with such a high speed into people, cars or maybe I could lost in definitelly...

Hi ata2k6, thank you for bringing this query to us. We do feel sad about what happened recently with your aircraft. If its convenient to you, may we ask for the serial number ( SN ) of the Mavic Air? I strongly suggest sending in the aircraft for assessment. Our data analyst team will conduct a thorough analysis of the incident. Kindly sync in your flight records and contact our Product Support to assist you in creating a case ticket. You may also send an email to support@dji.com. Do let us know if you have concerns or clarifications. Thank you.

Hi ata2k6, we're sorry to read the accident. Anyway, it's lucky to hear the unit has been recovered and unharmed, you may change the propeller and connect with DJI GO 4 to see if there is any error message. In case the drone needs to be serviced, you may start a case and send it in from here: https://www.dji.com/support If the unit is under warranty period, the designated team will arrange data analysis and offer you a proper solution according to the data analysis result, thank you!

Hello, started the drone home, without propellers, no errors were shown. I'm affraid to take off with props, until I find out what happened, to avoid possible incidents. Serial number is 0K1DF3F1BDVE58

I uploaded the blackbox file via Dji Assistant and reported to Dji, no response yet.
I don't want  to send the drone just to look at to Dji, i thought if somebody just take a look at the logfile could explain to me what exactly happened, how could the drone catch up to such a huge speed? And what to check at this point?
I did a firmware refresh and vision sensor calibration via DJI Assistant, did IMU and compass calibration. No errors shown up in app, engines started up fine without props, gimbal works correct...
It could be from cold air? It is possible to IMU module do lock up due to cold? I know MPU6050's need a constant temperature for correct measuring - such a huge speed is possible only at an extreme tilt angle, which should mean a wrong tilt measuring from IMU. And the speed was real, not just a wrong GPS read, it passed the walley like a bullet, those few hundred meters in a couple of seconds. I saw just before the flyaway a 0.15V cell deviation on battery, could this cause? It is a brand new battery with 2 or 3 cycles...  Too many questionable things...

Hello!
I analyzed via FRAP (Thanks, JJB!!!!) the logfile. It makes me confused. It looks like it was taken by a 80mph wind gust and smashed to the hillside, the drone tried to counteract, but it was in P-mode and got only about 14degrees tilt against that something, responded to controls, but it was too late. As a lesson, be careful when flying in walleys, even if there is absolutelly no wind at takeoff, situation can change dramatically at higher altitudes, and also can appear some serious gusts up there. As countermeasure, if I could hit Sport mode and climb, maybe i would get out of those turbulences and avoid smashing to hillside, and regain control, but in this particular case I had little to very little time to react. For me it was a cheap lesson, it costed me only a pair of props, and got a good gym climbing up on the hill, but it could be more expensive if drone would crash to rocky edge, at this speed it could surely be disintegrated into small bits, the foliage from the bushes and the thick leave cover on the ground saved him. Tough drone, anyway... wonder what would happen to a MP or MP2 gimbal when entered into the bush with 80 miles per hour.... This got not even a scratch on arms... I consider myself very -  very lucky.

ata2k6 Posted at 12-3 05:18
Hello!
I analyzed via FRAP (Thanks, JJB!!!!) the logfile. It makes me confused. It looks like it was taken by a 80mph wind gust and smashed to the hillside, the drone tried to counteract, but it was in P-mode and got only about 14degrees tilt against that something, responded to controls, but it was too late. As a lesson, be careful when flying in walleys, even if there is absolutelly no wind at takeoff, situation can change dramatically at higher altitudes, and also can appear some serious gusts up there. As countermeasure, if I could hit Sport mode and climb, maybe i would get out of those turbulences and avoid smashing to hillside, and regain control, but in this particular case I had little to very little time to react. For me it was a cheap lesson, it costed me only a pair of props, and got a good gym climbing up on the hill, but it could be more expensive if drone would crash to rocky edge, at this speed it could surely be disintegrated into small bits, the foliage from the bushes and the thick leave cover on the ground saved him. Tough drone, anyway... wonder what would happen to a MP or MP2 gimbal when entered into the bush with 80 miles per hour.... This got not even a scratch on arms... I consider myself very -  very lucky.

Hi Ata2K6,

wow, happy that you found your MA back!!

And yes, lots of wind. At the end loosing GPS reception to zero, so ATTI mode active.

cheers
JJB

JJBspark Posted at 12-4 12:16
Hi Ata2K6,

wow, happy that you found your MA back!!
[Image]

The interresting fact was, that down was absolutelly no wind at all, and on the footage also i didn't saw any tree movement. An 80 miles wind gust should be visible by picked up debris and I should anyway her the wind noise. So up to a certain percent it will remain a mistery the cause of this flyaway.
When the drone got to Atti mode it was already crashed in the bushes.

ata2k6 Posted at 12-5 04:47
The interresting fact was, that down was absolutelly no wind at all, and on the footage also i didn't saw any tree movement. An 80 miles wind gust should be visible by picked up debris and I should anyway her the wind noise. So up to a certain percent it will remain a mistery the cause of this flyaway.
When the drone got to Atti mode it was already crashed in the bushes.

Oke, so important question is than why did your drone drifted away and was not capable of holding position....

Speed indication doesn`t show correct in this situations, max about 22 m/s.

cheers
JJB

There are some battery/cell errors in the logs. What are the individual motor torque values? Could the cold weather have played a role?

BumblerBee Posted at 12-5 06:44
There are some battery/cell errors in the logs. What are the individual motor torque values? Could the cold weather have played a role?

I saw that too, it was a 0.15V cell voltage deviation. But this is a brand new, original Air battery, with 2 -3 cycles, it was fully charged.
The temperature was quite low, -7 degrees, but i did 2 more flights earlier that morning, without any problem. The problem is that I stopped recording when I saw the jerky movement, to not catch it on the footage, if I would let it on, it would bring us some very important informations.
Here is the video of the recording, with the very last moments right before the flyaway, you can see some vibratios, the linear yaw movement is made by me, that was the last command that I made to it.

JJBspark Posted at 12-5 05:06
Oke, so important question is than why did your drone drifted away and was not capable of holding position....

Speed indication doesn`t show correct in this situations, max about 22 m/s.

Yes, that was suspicious to me too, maybe still was an IMU error, what caused an extreme tilt?
A 130 kph (80 miles fast) wind, that can't be heared / seen is quite unplausable for me, you can see from the posted footage, that there is no wind at all, just look at the trees.
Anyway today or tomorrow mornig I will take it to the air, to see if there is any suspicious movement

ata2k6 Posted at 12-6 00:11
Yes, that was suspicious to me too, maybe still was an IMU error, what caused an extreme tilt?
A 130 kph (80 miles fast) wind, that can't be heared / seen is quite unplausable for me, you can see from the posted footage, that there is no wind at all, just look at the trees.
Anyway today or tomorrow mornig I will take it to the air, to see if there is any suspicious movement

Hi Ata2k6,

Good to see your video, what i see is familiar what i have experienced.
See my chart, the rapid yaw move is in the chart, what you see is ofcourse the heading excursions of the  gimbal with the cam. That`s different from the AC heading thus yaw moves. See my chart. Blue line is cam yaw, purple AC hawding and red line actual moving direction.

After that your MA starts to drift away.

My advice is to ask help from DJI, i think you have a "error" in your MA. Software related or hardware related?? DJI must explain this.

same happend here (but not the drifting) > https://forum.dji.com/thread-172298-1-1.html

cheers
JJB

I had a sudden strong wind.  Hit "pause" and then lowered it.  However it was not near as high wind as yours.

I rely on "AUSE" quite often

HendrikJB Posted at 12-6 03:17
I had a sudden strong wind.  Hit "pause" and then lowered it.  However it was not near as high wind as yours.

I rely on "PAUSE" quite often

It was only a few seconds in my case, after I figured out and tried evasive manuever (first climb, then climb and reverse, in the very last 2 seconds, but it was too late.) If I would climb from the first moment, I could avoid hillside, but on the other hand maybe higher was an even stronger wind, i had plenty of battery (had 88% when crashed) If it was only a spontaneous wind stroke, and if it was really just a wind gust and not an internal error, more probably I would regained control and land safely, i'm on FCC, full signal, so even if flyed a mile away I should have signal. But if it was an internal error, it means the drone would fly at that tremendous speed until battery drain and crash far far away in an unpopulated area

ata2k6 Posted at 12-6 04:41
It was only a few seconds in my case, after I figured out and tried evasive manuever (first climb, then climb and reverse, in the very last 2 seconds, but it was too late.) If I would climb from the first moment, I could avoid hillside, but on the other hand maybe higher was an even stronger wind, i had plenty of battery (had 88% when crashed) If it was only a spontaneous wind stroke, and if it was really just a wind gust and not an internal error, more probably I would regained control and land safely, i'm on FCC, full signal, so even if flyed a mile away I should have signal. But if it was an internal error, it means the drone would fly at that tremendous speed until battery drain and crash far far away in an unpopulated area

Oh OK. I never considered that scenario. Just happy that you only lost the propellers.

I did a refresh on the .500 FW, via Assistant, visual sensor calibration too  put Air few minutes to fridge, fresh battery in, IMU calibration, got outdoors, made a compass calibration.
Now it seems to fly like it supposed, no yaw, no shake, made a 2 km and a 6.5km long flight, flawles.. If it will remain as is... it wood be lovely, if not, I'll sell it.

I did find it strange that no wind warnings came up..

Deminic Posted at 12-11 02:31
I did find it strange that no wind warnings came up..

As I said earlier, this flyaway remains a mistery.
Offcourse DJI will not take a look on this situation, despite the fact that I sent the blackbox content from DJ Assistant to them with the exact description of situation, and they also have acces upon the logfile here in the topic, I will not send my drone in because I'm convinced that this issue can be sorted out remotelly too.

I had a similar experience with my MA, the only difference was temperature, I was flying in +4 deg. temperature. in my case, i could not calibrate downward vision sensors after crash. so i created a case. DJI service was good, although i did not get any reasoning,  I got a replacement drone because of hardware failure.

Here is the post
https://forum.dji.com/thread-172641-1-1.html