Game of Drones - Mavic crash

[MML]

This might be the first recorded Mavic crash in Westeros!

As many of you who watch Game of Thrones will know, a lot of filming is done in Northern Ireland. At 6am yesterday I set off to the Dark Hedges (the location of the Kingsroad in the show) to meet a friend and get some pics and video. I arrived at sun-up and flew through 2 batteries. The Mavic performed 100% perfectly.

After we left, we went to Ballintoy Harbour (The Iron Islands in the show) to shoot some more. Everything was going normally, at one point I took off to fly to the harbour mouth and return to home point, approx 250m round trip, shooting birds-eye-view. This did not go according to plan.

The first thing I noticed was the aircraft yawing as I was was flying in a straight line. I tried to keep it straight but it wouldn't play ball. Then it started drifting around, I was fighting to recover it as best I could. I tried going into sport mode and even RTH at one point to regain control. The rest is a blur to be honest; then it smacked into this cliff face:

Google Maps

Upon inspection, the front arms had been pushed backwards, all the props bar one were broken, but the weird thing is, there were no signs of major impact marks like cracks/chips etc. The gimbal was folded back as in the photo but seems undamaged.
One rear arm feels a bit wrong when folded up, as if there's too much play in it.

I've since started up the Mavic, the camera does it's little warmup dance, and seems to operate normally. All motors (without props) seemed to run as normal.

I didn't notice any warnings about compass errors etc. I did see a brief mention of interference to the controller, but I moved away from the car I was standing next to and the warning disappeared.

I've uploaded the last videos to Youtube:

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

You can see how it starts to rotate, and how it starts drifting off.
Also attached is the last photo I took and some shots of the damage.

Here are the logs, I think they're the correct ones.

Dropbox - DJI_ASSISTANT_EXPORT_FILE_2017-10-28_19-35-34.DAT

Dropbox - DJIFlightRecord_2017-10-27_[10-52-05].txt

If anyone can analyse the logs and let me know what happened I'd be very grateful!

 

[BudWalker]

This might be the first recorded Mavic crash in Westeros!

As many of you who watch Game of Thrones will know, a lot of filming is done in Northern Ireland. At 6am yesterday I set off to the Dark Hedges (the location of the Kingsroad in the show) to meet a friend and get some pics and video. I arrived at sun-up and flew through 2 batteries. The Mavic performed 100% perfectly.

After we left, we went to Ballintoy Harbour (The Iron Islands in the show) to shoot some more. Everything was going normally, at one point I took off to fly to the harbour mouth and return to home point, approx 250m round trip, shooting birds-eye-view. This did not go according to plan.

The first thing I noticed was the aircraft yawing as I was was flying in a straight line. I tried to keep it straight but it wouldn't play ball. Then it started drifting around, I was fighting to recover it as best I could. I tried going into sport mode and even RTH at one point to regain control. The rest is a blur to be honest; then it smacked into this cliff face:

Google Maps

Upon inspection, the front arms had been pushed backwards, all the props bar one were broken, but the weird thing is, there were no signs of major impact marks like cracks/chips etc. The gimbal was folded back as in the photo but seems undamaged.
One rear arm feels a bit wrong when folded up, as if there's too much play in it.

I've since started up the Mavic, the camera does it's little warmup dance, and seems to operate normally. All motors (without props) seemed to run as normal.

I didn't notice any warnings about compass errors etc. I did see a brief mention of interference to the controller, but I moved away from the car I was standing next to and the warning disappeared.

I've uploaded the last videos to Youtube:

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

You can see how it starts to rotate, and how it starts drifting off.
Also attached is the last photo I took and some shots of the damage.

Here are the logs, I think they're the correct ones.
Dropbox - DJI_ASSISTANT_EXPORT_FILE_2017-10-27_21-32-10.DAT

Dropbox - DJIFlightRecord_2017-10-27_[10-52-05].txt

If anyone can analyse the logs and let me know what happened I'd be very grateful!

The .DAT you submitted isn't the correct one. The length of this .DAT is about 94 secs but there is no motorStart. This .DAT appears to have been created 2017-10-27 20:21:42 GMT. Take a look here for info on retrieving the right .DAT. Could you try again please.

I did look at the .txt and appears that there was GPS issue that may have contributed to this incident.

 

[sar104]

The sequence of events seems to be clear enough. At 816 seconds the aircraft went into ATTI mode (reason unknown) and started to drift and yaw, as expected. It continued to respond to stick inputs but it appears that you were unable to control it in that mode - something that does require practice. At 825 seconds it looks like it hit the cliff/slope and tumbled.

 

[BudWalker]

The sequence of events seems to be clear enough. At 816 seconds the aircraft went into ATTI mode (reason unknown) and started to drift and yaw, as expected. It continued to respond to stick inputs but it appears that you were unable to control it in that mode - something that does require practice. At 825 seconds it looks like it hit the cliff/slope and tumbled.

View attachment 23892

The reason that it went into ATTI at 763 and 816 was that gpsHealth dropped from 5 to 1. Maybe the .DAT can provide some insight as to why this happened.

 

[sar104]

The reason that it went into ATTI at 763 and 816 was that gpsHealth dropped from 5 to 1. Maybe the .DAT can provide some insight as to why this happened.
View attachment 23894

I guessed that would be the case but the txt interpreter that I'm using doesn't show GPS health. I need to put CsvView on a windows box.

 

[MML]

Yikes... let me see if I can get the correct one...

 

[MML]

I think this is the right one... sorry about that!

Dropbox - DJI_ASSISTANT_EXPORT_FILE_2017-10-28_19-35-34.DAT

 

[sar104]

I think this is the right one... sorry about that!

Dropbox - DJI_ASSISTANT_EXPORT_FILE_2017-10-28_19-35-34.DAT

No - that's the same file that you linked before.

 

[MML]

No - that's the same file that you linked before.

Not sure what happened there... I was sure that was the correct one from the aircraft.
It has to be one of these two, as the times match the time of the crash, see this screenshot from DJI Assistant. I've just re-downloaded them off the drone.

Links to both:
Dropbox - DJI_ASSISTANT_EXPORT_FILE_2017-10-28_20-56-44.DAT

Dropbox - DJI_ASSISTANT_EXPORT_FILE_2017-10-28_21-00-59.DAT

 

[sar104]

Not sure what happened there... I was sure that was the correct one from the aircraft.
It has to be one of these two, as the times match the time of the crash, see this screenshot from DJI Assistant. I've just re-downloaded them off the drone.

Links to both:
Dropbox - DJI_ASSISTANT_EXPORT_FILE_2017-10-28_20-56-44.DAT

Dropbox - DJI_ASSISTANT_EXPORT_FILE_2017-10-28_21-00-59.DAT

The second file was the correct one, containing FLY288.DAT.

 

[Callum]

Had my mavic a month and today got my first taste of atti mode.

Went into a bit of a panic like yourself then gained some height and gave myself some breathing room and time to work it out.

Wish there was an atti switch on mavic so new pilots could practice it before ending up in a situation like yourself.

 

[MML]

Had my mavic a month and today got my first taste of atti mode.

Went into a bit of a panic like yourself then gained some height and gave myself some breathing room and time to work it out.

Wish there was an atti switch on mavic so new pilots could practice it before ending up in a situation like yourself.

Yeah, the panic certainly hits you when you realise you have a very expensive piece of hardware rapidly going out of control over water! The battery warning beep doesn't help you remain calm either.

I put a spare set of props on today and tried a test flight. I did a compass calibration and a camera calibration before takeoff.
I went through a full battery's worth of flight time and all seems to be fine, there's full unobstructed range of motion on the camera. I'm concerned about the play in the rear arm joint though.

Here is a quick video:

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

 

[sar104]

The switch into ATTI seems to be related to large discrepancies between the yaw values computed by the FC from the IMU sensor fusion data and the yaw based on magnetometer data alone. That has nothing to do with GPS per se, but often seems to trigger a switch to ATTI.

 

[MML]

Thanks for the assessment, it’s great to have some light shed on this!

So is this something that can’t really be prevented or avoided? Or is it something that can be fixed with a software update?

To be honest, having this potentially happen again is something that puts me off flying a bit. The fear is always there.

Thinking back, this is the third time I recall it switching modes at random. I posted about the first incident about 5 months ago; the second was last month, but the GPS mode came back after about 10 seconds.

 

[sar104]

Thanks for the assessment, it’s great to have some light shed on this!

So is this something that can’t really be prevented or avoided? Or is it something that can be fixed with a software update?

To be honest, having this potentially happen again is something that puts me off flying a bit. The fear is always there.

Thinking back, this is the third time I recall it switching modes at random. I posted about the first incident about 5 months ago; the second was last month, but the GPS mode came back after about 10 seconds.

I would need to take a much closer look at the magnetometer data, which I suspect is the cause of the problem. The two magnetometers show significant differences towards the end of the flight that I don't think are normal. You might want to try a compass calibration, even if the FC is not requesting it.

 

[BudWalker]

Not sure what happened there... I was sure that was the correct one from the aircraft.
It has to be one of these two, as the times match the time of the crash, see this screenshot from DJI Assistant. I've just re-downloaded them off the drone.

Links to both:
Dropbox - DJI_ASSISTANT_EXPORT_FILE_2017-10-28_20-56-44.DAT

Dropbox - DJI_ASSISTANT_EXPORT_FILE_2017-10-28_21-00-59.DAT

I took a look at FLY288 which contained 4 flights - the incident you described was in the last one. In summary, the Mavic had switched from compass # 1 to compass #2 which wasn't working very well. When the RTH was initiated the Mavic decided GPS+ATTI couldn't be used and switched to ATTI.

At the end of flight 2 the Mavic had been using compass #1 (back compass) which it normally uses. It then switched to compass #2 (front compass). This can be seen by looking at the both magZ:1 and magZ:2 and comparing to magZ. The switch occurs at time 412.51

Also, from the eventLog stream,
414.672 :[L-FDI]NS(0) COMPASS(0): fault on , stuck
414.672 : [L-FMU/DM]Busy Device Changed. Type:compass1, <ID:3 idx:0-->ID:4 idx:1>, change times:

Following the compass switch there were no anomalies until the 4th flight at time 803. For some reason the mag:X(2) bias shifted which was the cause of the Yaw/magYaw separations mentioned by @sar104 .



As can be seen the Mavic managed to deal with the anomaly without complaint. It was only when RTH was initiated that the Mavic decided that GPS+ATTI was not possible. This from the eventLog stream.
881.713: [L-FLYMODE][ACT.msg] ACT_GO_HOME gps level no good enough for go home!

EDIT: @sar104 I guess we cross posted

 

[sar104]

I took a look at FLY288 which contained 4 flights - the incident you described was in the last one. In summary, the Mavic had switched from compass # 1 to compass #2 which wasn't working very well. When the RTH was initiated the Mavic decided GPS+ATTI couldn't be used and switched to ATTI.

At the end of flight 2 the Mavic had been using compass #1 (back compass) which it normally uses. It then switched to compass #2 (front compass). This can be seen by looking at the both magZ:1 and magZ:2 and comparing to magZ. The switch occurs at time 412.51
View attachment 24078
Also, from the eventLog stream,
414.672 :[L-FDI]NS(0) COMPASS(0): fault on , stuck
414.672 : [L-FMU/DM]Busy Device Changed. Type:compass1, <ID:3 idx:0-->ID:4 idx:1>, change times:

Following the compass switch there were no anomalies until the 4th flight at time 803. For some reason the mag:X(2) bias shifted which was the cause of the Yaw/magYaw separations mentioned by @sar104 .
View attachment 24081
View attachment 24082

As can be seen the Mavic managed to deal with the anomaly without complaint. It was only when RTH was initiated that the Mavic decided that GPS+ATTI was not possible. This from the eventLog stream.
881.713: [L-FLYMODE][ACT.msg] ACT_GO_HOME gps level no good enough for go home!

EDIT: @sar104 I guess we cross posted

Nicely explained. I was also puzzled why the discrepancies were ignored to begin with, but perhaps it has stricter criteria for RTH - that would make sense. Do you think that calibration is indicated?

 

[BudWalker]

...
Do you think that calibration is indicated?

I wouldn't think so, but, then, I don't have another suggestion. Compass #2 was fine throughout flight 3 and part of flight 4. It was only towards the end where the discrepancy arose. If it were a calibration issue I would expect to see the problem starting with the switch to compass #2. But, I'm stumped....

 

[sar104]

I wouldn't think so, but, then, I don't have another suggestion. Compass #2 was fine throughout flight 3 and part of flight 4. It was only towards the end where the discrepancy arose. If it were a calibration issue I would expect to see the problem starting with the switch to compass #2. But, I'm stumped....

If I recall correctly we saw a similar situation in a flight record a month or so back - the sudden appearance of a discrepancy like that mid-flight.

 

[BudWalker]

If I recall correctly we saw a similar situation in a flight record a month or so back - the sudden appearance of a discrepancy like that mid-flight.

Yes, but this is even more interesting. @MML had submitted a FLY131.DAT back in June. Pretty much the same problem. The eventLog says that compass #1 is stuck and the flight uses compass #2. Towards the of the flight problems with Mag:X(2) get large enough that a switch to ATTI occurs.


Here's the Dropbox link for FLY131
Dropbox - FLY131.DAT

The compass had been calibrated at the beginning of FLY131. And, it's also been calibrated since FLY131.