Mavic Air took off by itself ascended to 400m and then crashed - no input

[Phil Collins]

Hi all –

Very new user here. My apologies for not following typical standards – I must admit I’m still a bit flustered.

I just unboxed my Mavic Air and after bringing the battery to a full charge, I took it outside, after updating all various apps and firmwares etc… and because it was the first time running it the app made me do simulator mode, which I obliged, because… why not? I fly a bit in the app and then touch restart and suddenly the props on my drone (which is on the ground about 3 metres away from me) start spinning up and it launches straight up into the air at full speed. It had to go up 400 metres, if I was guessing. As it soared upward – which I was not expecting; I was expecting it to hover at 1.2m, even though I didn’t actively want it to start. As it climbed, at what I’d guess was 200m, I hit the RTH button, which seemed to have no effect as the drone moved forward about 100m and then plummeted to the ground – with props still spinning the entire time. To be clear, I never even touched the controls/sticks once the props started turning as I was shocked they were doing so. The only input I made was the RTH button.

Fortunately, it landed in an empty lot – make of concrete, of course – and is all busted up from the hard impact.

Obviously it goes without saying that this could have been very bad, and that I am rather lucky right now.

I have checked the app and there is no record of a flight. No log. No nothing.

I’ve already logged the incident with DJI via the app, and will call first thing in the morning once customer support is open – but I am slightly shaken by what could’ve happened.

Has anyone else experienced anything like this?

Thanks!

 

[ABHIII]

Ghost Mavic Air

 

[AnzacJack]

With no logs it’s going to be pretty hard to work out what happened

 

[Phil Collins]

With no logs it’s going to be pretty hard to work out what happened

That's very true.

Is it possible there's a log on the actual device and it simply didn't register on the app?

 

[Kilrah]

There are logs on the aircraft but only DJI can read them.
They likely have a serious bug with simulator mode.

 

[AMann]

From Meta4...

To see what the data shows ....

DJI Flight Log Viewer | Phantom Help

Upload and view flight logs from your DJI drones

www.phantomhelp.com

Follow the instructions there to upload your flight record from your phone or tablet. That will give you a detailed report on the flight data. Come back and post a link to the report it gives you.

 

[White Swan]

 

[DDS]

Hi all –

Very new user here. My apologies for not following typical standards – I must admit I’m still a bit flustered.

I just unboxed my Mavic Air and after bringing the battery to a full charge, I took it outside, after updating all various apps and firmwares etc… and because it was the first time running it the app made me do simulator mode, which I obliged, because… why not? I fly a bit in the app and then touch restart and suddenly the props on my drone (which is on the ground about 3 metres away from me) start spinning up and it launches straight up into the air at full speed. It had to go up 400 metres, if I was guessing. As it soared upward – which I was not expecting; I was expecting it to hover at 1.2m, even though I didn’t actively want it to start. As it climbed, at what I’d guess was 200m, I hit the RTH button, which seemed to have no effect as the drone moved forward about 100m and then plummeted to the ground – with props still spinning the entire time. To be clear, I never even touched the controls/sticks once the props started turning as I was shocked they were doing so. The only input I made was the RTH button.

Fortunately, it landed in an empty lot – make of concrete, of course – and is all busted up from the hard impact.

Obviously it goes without saying that this could have been very bad, and that I am rather lucky right now.

I have checked the app and there is no record of a flight. No log. No nothing.

I’ve already logged the incident with DJI via the app, and will call first thing in the morning once customer support is open – but I am slightly shaken by what could’ve happened.

Has anyone else experienced anything like this?

Thanks!

Please post the results of the cause.
We would be interested to know what happened. That’s strange behavior.

 

[Phil Collins]

From Meta4...

To see what the data shows ....

DJI Flight Log Viewer | Phantom Help

Upload and view flight logs from your DJI drones

www.phantomhelp.com

Follow the instructions there to upload your flight record from your phone or tablet. That will give you a detailed report on the flight data. Come back and post a link to the report it gives you.

I think I've been able to do this. There were only .DAT files on the device - so I used DatCon to convert and of the two files, these are what came back:

DJI Flight Log Viewer - PhantomHelp.com

DJI Flight Log Viewer - PhantomHelp.com

It also gave me some .csv files and .kml files. These links look pretty bare to me - but any insights would be glorious.

 

[AMann]

I think I've been able to do this. There were only .DAT files on the device - so I used DatCon to convert and of the two files, these are what came back:

DJI Flight Log Viewer - PhantomHelp.com

DJI Flight Log Viewer - PhantomHelp.com

It also gave me some .csv files and .kml files. These links look pretty bare to me - but any insights would be glorious.

No data shows up on your links because it needs the TXT files, not the DAT files. Try again, the directions are on the site.

 

[BudWalker]

I think I've been able to do this. There were only .DAT files on the device - so I used DatCon to convert and of the two files, these are what came back:

DJI Flight Log Viewer - PhantomHelp.com

DJI Flight Log Viewer - PhantomHelp.com

It also gave me some .csv files and .kml files. These links look pretty bare to me - but any insights would be glorious.

The .txt files produced by DatCon are not the .txt files being requested here. The .txt files being requested here should be in the FlightRecords folder. Take a look at the last part of this page for more info on how to do this.
DJI Flight Log Viewer | Phantom Help

The .csv files you produced using DatCon are where the real action is. Better yet, why don't you just provide the .DAT files so others can take a look.

 

[Phil Collins]

Here's the DAT files - I'll re-look at the link - but when I go into my Android files - this is all I see:


I then click through MCDatFlightRecords and this is what shows up:


And I've attached those two .DAT files.

Appreciate any insight - I've logged a repair request with DJI and am awaiting the mailing slip so they can look and analyse as well.

 

[sar104]

If FLY022.DAT reflects real data then I've never seen anything quite like that before:



It does appear to show an auto-takeoff command at 57 seconds while in simulator mode, leading to a motor start and then an actual takeoff at 68 seconds. Then control failed - it climbed to around 126 meters and then the FC set the motors back to idle, which didn't go well as you might expect, leading to an uncontrolled descent reaching 14 m/s. At 93 seconds CSC was armed due to control failure and executed by sticks down and out, with the aircraft still around 30 meters up.

It's really hard to decipher what is real and what is imaginary in that log, but the GPS data appear to be real, and you do appear to have provided some stick inputs, including the final CSC Were you here, and do those data correspond with your recollection?





57.730 : 12900 [L-API]get_id_index_by_basic_cmd: cmd 1 index 0
57.730 : 12900 [L-CMD]exec_fc_basic_function index 0
57.730 : 12900 [L-API]auto_take_off_check
57.730 : 12900 [L-API]basic_cmd_auto_take_off
57.730 : 12900 [L-N_MIS]Init plan 1 from requester 6 for reason 14. Pre plan 5
57.730 : 12900 [L-TAKEOFF]enter auto takeoff plan
57.730 : 12900 [L-N_MIS]Exit mis 3. New mis 4
57.730 : 12900 [L-N_MIS]Switch new mis 4 req 6 from mis 3 req 2
57.731 : 12900 [L-FMU/MOTOR]Simultor passed, but may fail in real for reason 74: ns abnormal
57.731 : 12900 [L-FMU/MOTOR]Start. reason:ACT.Takeoff
57.731 : 12900 [L-PWM]set actuator mode:CTRL
57.731 : 12900 [L-FDI]tilt when motor on = 0.156213, static flag = 0
57.732 : 12900 [L-FMU/FSM]state changed. cur: on ground
57.737 : 12900 [L-FMU/FSM]state changed. cur: in air
57.751 : 12901 [L-FDI][CTRL]: fault on , height_ctrl_fail
57.770 : 12902 [L-RC]craft ctrl failed!!!
58.927 : 12960 [L-FDI][sensor_api_info] head(2,0,31,4).|
59.236 : 12975 [L-N_MIS]Exit mis 4. New mis 6
59.236 : 12975 [L-N_MIS]Switch new mis 6 req 6 from mis 4 req 6
59.246 : 12976 [L-N_MIS]req abort all for reach target height
59.251 : 12976 [L-TAKEOFF]exit auto takeoff plan!
59.251 : 12976 [L-N_MIS]Switch new mis 3 req 2 from mis 26 req 16
59.251 : 12976 [L-FLYMODE]SWITCH2HOVER!
59.784 : 13003 [L-RC]craft ctrl failed!!!
61.800 : 13104 [L-RC]craft ctrl failed!!!
63.814 : 13205 [L-RC]craft ctrl failed!!!
64.285 : 13228 [L-CFG]PM:[Mobile APP] "novice_func_enabled" to 0||
64.435 : 13236 [L-FDI][FDI] ns atti(0) heading is outlier
64.435 : 13236 [L-FDI][FDI] ns atti(1) heading is outlier
64.452 : 13237 [L-FDI]ns req:fdi,0to1,reason:disagree_yaw,result:fail for disagree_yaw
65.729 : 13300 [L-BATTERY]get_cell_voltage_callback_ack failed!||
65.829 : 13306 [L-RC]craft ctrl failed!!!
67.843 : 13407 [L-RC]craft ctrl failed!!!
67.866 : 13408 [L-FDI][FDI] ns atti(0) alti is outlier
67.866 : 13408 [L-FDI][FDI] ns atti(1) alti is outlier
67.996 : 13414 [L-FDI][FDI] ns atti(0) vgz is outlier
67.996 : 13414 [L-FDI][FDI] ns atti(1) vgz is outlier
68.115 : 13420 [L-FMU/FSM]near ground
68.624 : 13446 [L-FMU/FSM]not near ground
69.858 : 13508 [L-RC]craft ctrl failed!!!
71.874 : 13609 [L-RC]craft ctrl failed!!!
73.888 : 13710 [L-RC]craft ctrl failed!!!
74.885 : 13760 [L-FDI][sensor_api_info] head(2,0,31,4).|
75.903 : 13811 [L-RC]craft ctrl failed!!!
76.063 : 13819 [L-RC]rc cmd:STD_GO_HOME
76.067 : 13819 [L-N_MIS]Init plan 0 from requester 9 for reason 8. Pre plan 5
76.077 : 13819 [L-N_MIS]Exit mis 3. New mis 11
76.077 : 13819 [L-N_MIS]Switch new mis 11 req 9 from mis 3 req 2
76.861 : 13859 [L-RC]Start landing. rc vib!
77.917 : 13912 [L-RC]craft ctrl failed!!!
79.932 : 14013 [L-RC]craft ctrl failed!!!
81.948 : 14114 [L-RC]craft ctrl failed!!!
82.927 : 14163 [L-FDI][FDI] ns atti(0) alti is outlier
82.927 : 14163 [L-FDI][FDI] ns atti(1) alti is outlier
82.977 : 14165 [L-FDI][FDI] ns atti(0) alti is outlier
82.977 : 14165 [L-FDI][FDI] ns atti(1) alti is outlier
83.027 : 14168 [L-FDI][FDI] ns atti(0) alti is outlier
83.027 : 14168 [L-FDI][FDI] ns atti(1) alti is outlier
83.106 : 14172 [L-FDI][FDI] ns atti(0) vgz is outlier
83.106 : 14172 [L-FDI][FDI] ns atti(1) vgz is outlier
83.961 : 14215 [L-RC]craft ctrl failed!!!
85.977 : 14316 [L-RC]craft ctrl failed!!!
90.006 : 14518 [L-RC]craft ctrl failed!!!
90.844 : 14560 [L-FDI][sensor_api_info] head(2,0,31,4).|
92.021 : 14619 [L-RC]craft ctrl failed!!!
93.498 : 14693 [L-FMU/LED]action changed. forearm LED outer ctrl off:Ctrl forearm LED off by cam(2)
93.697 : 14703 [L-RC]arm stop because of craft fail in near history
93.697 : 14703 [L-RC]rc cmd:STD_STOP_MOTOR
93.697 : 14703 [L-FMU/LED]action changed. green(0)
93.716 : 14704 [L-FMU/MOTOR] Stop. reason:MR_REQ_RC_NORMAL
93.716 : 14704 [L-FMU/MOTOR]Total start times: 2, time: 66.08
93.716 : 14704 [L-N_MIS]GoHome Check Failed: motor off!
93.716 : 14704 [L-N_MIS]Check fail of plan 0 from requester 9 for reason 22
93.716 : 14704 [L-N_MIS]Exit plan 0
93.716 : 14704 [L-N_MIS]Fail to check mis 11. Reset
93.716 : 14704 [L-N_MIS]Exit mis 11. New mis 26
93.716 : 14704 [L-N_MIS]reset_smart_landing_status by: exit landing
93.716 : 14704 [L-N_MIS]Switch new mis 26 req 16 from mis 11 req 9
93.717 : 14704 [L-N_MIS]req abort all for motor stopped
93.717 : 14704 [L-FDI][CTRL]: fault off, height_ctrl_fail
93.719 : 14704 [L-FMU/FSM]state changed. cur: motor off

 

[Phil Collins]

Wow - I have very little clue what I'm looking at - but it seems like pretty wild information to me.

I feel adamant that there was no input - except for the RTH button press, which doesn't seem to logged as pressed. I suppose I may have touched a stick to see if I could control movement - but I certainly didn't initiate the emergency stop procedure, that I know for sure. So, that's interesting.

My calculation of distance aside (eg: 400m - which, it certainly seemed that high to me) - this seems to be a relatively accurate recount of what happened, with the exception of user input. There just simply wasn't any.

The image you sent is where takeoff occurred. The red 'X' is where the Air ate it:



Has anyone ever experienced anything like this? I am fearful I did something here, but also feel confident that this is an anomaly - albeit a concerning one - and that I now need to worry about AI trying to kill us all. Is this the beginning of SkyNet?

 

[Phil Collins]

Also @sar104 - thanks! Appreciate you.

 

[sar104]

Wow - I have very little clue what I'm looking at - but it seems like pretty wild information to me.

I feel adamant that there was no input - except for the RTH button press, which doesn't seem to logged as pressed. I suppose I may have touched a stick to see if I could control movement - but I certainly didn't initiate the emergency stop procedure, that I know for sure. So, that's interesting.

My calculation of distance aside (eg: 400m - which, it certainly seemed that high to me) - this seems to be a relatively accurate recount of what happened, with the exception of user input. There just simply wasn't any.

The image you sent is where takeoff occurred. The red 'X' is where the Air ate it:

Has anyone ever experienced anything like this? I am fearful I did something here, but also feel confident that this is an anomaly - albeit a concerning one - and that I now need to worry about AI trying to kill us all. Is this the beginning of SkyNet?

These are the stick inputs recorded over the duration that the aircraft was actually flying. They don't appear to have had any effect on the aircraft. So all other inputs aside - did you definitely not execute a CSC just before the crash?



The RTH command was logged in the event stream at 76 seconds:

76.063 : 13819 [L-RC]rc cmd:STD_GO_HOME

 

[Phil Collins]

Thanks Sar - I've just spent time reading and wrapping my head around the data. I can now see the RTH input - and a lot of failed ctrl readings too.

I firmly believe any input after 72 seconds wasn't me - although in all the confusion and overall lack of control of this weird situation, it's totally plausible I am simply not recalling trying to take control. However, I can say with absolute confidence that I did not initiate the CSC - unless it was a subconscious reaction, which I find highly unlikely as the drone had moved out of direct line of sight by that point and killing the engines blind doesn't seem a good idea. Easy to say in hindsight here, of course. But I still feel confident I did not initiate that command.

 

[AnzacJack]

I’m not a fan of being connected to the bird while in sim mode. It just seems to be taking unnecessary risk of just this type of thing. But I guess this is prob a very rare occupancy so unlikely DJI would make any changes

 

[Phil Collins]

I’m not a fan of being connected to the bird while in sim mode. It just seems to be taking unnecessary risk of just this type of thing. But I guess this is prob a very rare occupancy so unlikely DJI would make any changes

Personally I was frustrated as it seemed forced upon me - but I obliged being that it was the first time flying this drone and I wanted to check DJI's internal boxes. I naturally presumed that a simulation would be just that - I can tell you I haven't been that shocked in a while.

 

[sar104]

The flight track looks like this: