I screwed up - however

[KelloggKid]

Let me start by saying I messed up and made several mistakes. I had reached the experience level where I was probably a little cocky, hadn’t flown in several weeks, and made a hasty decision to try and grab a quick shot as the lighting was impressive.

Our house is on a ridge with the back of the house overlooking a steep, heavily forested slope leading to a picturesque river. Normally, if I were to fly out toward the river, I would set the MA out in front of our house on the paved driveway, make sure I had at least 10 satellites locked in and the Home Point established. Then I would fly up and over the house and let it hover while I walked through the house and out onto our back deck from which I would control the remainder of the flight.

This time I was in a huge hurry and took off instead from the deck itself with only 5 satellites and no HP set. The flight lasted a total of 58 seconds. At the 37 second point the MA had 10 satellites and at 42 seconds it set a Home Point. While this was not the location from which I had taken off, it was about 15’ higher and only out about 60’ so I didn’t worry about it. At that point in time I looked away from the MA just long enough to glance at the controller to make certain it was in photo mode not video mode. When I looked back up the drone was gone and the sound of the props had ceased. At 55.2 seconds a Motor Blocked was recorded.

The drone appears to be in to top of a tree about 100’ off the ground. We have not been able to located from down below and a neighbor ran through two battery packs on his Mavic 2 trying unsuccessfully to spot it from above.

I realize I messed up big time but am at a loss as to why the drone was right in front of me one minute and gone less than a second or two later. I am planning on replacing the drone and would really like to know what went wrong so would appreciate any thoughts on the matter.

 

[BigAl07]

To be honest, one of my sUAS was in pretty poor shape after doing almost exactly what you did. It was a Phantom3Pro and I had it hovering below some tree limbs (maybe 5'+) and I looked down at the Transmitter to confirm camera settings (we shoot in AEB) and as I glanced down I heard the impact of the aircraft to leaves and looked up to see it fighting the limbs (and losing) and then a whirling sound as one motor became stuck and the aircraft tumbled about 50' to terra firma.

Lessons learned:

1) When I need to look at the transmitter I lift it up into my line of sight so I am only taking eyes of the aircraft for a split second
2) there is a bit of drift (more in older aircraft than newer) when hovering especially when near objects
3) Trees are DRONE MAGNETS!!

 

[DanMan32]

My P3 tended to screech as if in pain when it hit something and tumbled down. What a horrible sound.

 

[VouilEz]

It happens to the best mate, and it;s refreshing to hear a Experienced Flyer say it's happened to them.
Though i fear his empathy, honesty comes from the fact it's such an old model.

 

[FoxhallGH]

Let me start by saying I messed up and made several mistakes. I had reached the experience level where I was probably a little cocky, hadn’t flown in several weeks, and made a hasty decision to try and grab a quick shot as the lighting was impressive.

Our house is on a ridge with the back of the house overlooking a steep, heavily forested slope leading to a picturesque river. Normally, if I were to fly out toward the river, I would set the MA out in front of our house on the paved driveway, make sure I had at least 10 satellites locked in and the Home Point established. Then I would fly up and over the house and let it hover while I walked through the house and out onto our back deck from which I would control the remainder of the flight.

This time I was in a huge hurry and took off instead from the deck itself with only 5 satellites and no HP set. The flight lasted a total of 58 seconds. At the 37 second point the MA had 10 satellites and at 42 seconds it set a Home Point. While this was not the location from which I had taken off, it was about 15’ higher and only out about 60’ so I didn’t worry about it. At that point in time I looked away from the MA just long enough to glance at the controller to make certain it was in photo mode not video mode. When I looked back up the drone was gone and the sound of the props had ceased. At 55.2 seconds a Motor Blocked was recorded.

The drone appears to be in to top of a tree about 100’ off the ground. We have not been able to located from down below and a neighbor ran through two battery packs on his Mavic 2 trying unsuccessfully to spot it from above.

I realize I messed up big time but am at a loss as to why the drone was right in front of me one minute and gone less than a second or two later. I am planning on replacing the drone and would really like to know what went wrong so would appreciate any thoughts on the matter.

We have seen posts a bit like this before. You'd need to have the logs looked over by somebody with more expertise than me, but my guess is that the reason for the sudden change in heading, is all down to the initial haste of the launch! If you have not let the Mavic have time to find its bearings, then what can happen is that it will receive all the data to do so after it has set off on the flight. At that time, there may be a difference that occurs due to the Mavic suddenly 'thinking' it is flying a course that it's not! The Mavic then compensates, but that actually makes it head off in an entirely new direction (and attitude), with control inputs no longer being sensible. I hope you get your Mavic Air back ... but for future reference, this sounds like a case of 'Haste makes Waste!' ...

 

[Dakarenduro]

Probably too late now but you could have initiated the flashing lights and beeper to help locate it...

 

[KateFly]

Probably too late now but you could have initiated the flashing lights and beeper to help locate it...

How do you activate those please?

 

[Dakarenduro]

How do you activate those please?

From Go4 app you select menu, select find my drone, on map press the red drone arrow, then as long as you are still connected to aircraft you can select the option.

 

[KelloggKid]

Probably too late now but you could have initiated the flashing lights and beeper to help locate it...

I knew that the DJI Go4 app had a Find My Drone feature but didn't realize that it had the lights and beeper capability until after my battery had died. Wish I had know that as it might have allowed me to determine which tree it is in.

I'd appreciate it if someone with more experience than I could review the logs and tell me what they think happened. At the time I looked away the drone had locked on 15 satellites, was in a stable hover, and I issued no stick inputs. The very last entry in the flight record refers to a compass error so perhaps that may have been a factor. Obviously the biggest factor was my initial stupidity but beyond that I'd like to know what caused the drone to wonder off on it's own. I might add there was no wind and the drone sat in a stable hover for about 9-10 seconds after it had acquired the 15 satellites and set the HP. When it was hovering it was 25-30' in all directions away from any trees. From the point where it was hovering, and during that brief period when I looked down at the controller, it spun 180 degrees and appears to have moved laterally about 75'.

 

[BigAl07]

From Go4 app you select menu, select find my drone, on map press the red drone arrow, then as long as you are still connected to aircraft you can select the option.
View attachment 77250View attachment 77250View attachment 77251View attachment 77252
View attachment 77250
View attachment 77251
View attachment 77252

And I learned something new today!! I need to try this out and get familiar with it.. Just-In-Case

 

[synaesthesia]

And I learned something new today!! I need to try this out and get familiar with it.. Just-In-Case

Indeed, every day is a school day. I had no idea it could even do this... super useful feature.

 

[BobLaQ]

From Go4 app you select menu, select find my drone, on map press the red drone arrow, then as long as you are still connected to aircraft you can select the option.

View attachment 77250
View attachment 77251
View attachment 77252

So cool - thank you for that. So much to learn, so little time...

 

[surfingjoe]

I'm so glad I've read this thread. I've not lost a drone, hope I never will. if I do, I now know of a feature I didn't know exists in the app. Thank You Dakarenduro, for being the detailed instructions.

 

[sar104]

This is a slightly strange one. The flight was basically as you described - you took off without a GPS lock in OPTI mode. At 42 seconds the home point was set when the navHealth level reached 4/5:



The anomalous behavior started at 49 seconds, when the FC added forward pitch and left roll without any stick input.





As a result the aircraft the aircraft moved 9 meters to the southeast and collided with a tree at 54 seconds:



At 57 seconds the FC determined that the aircraft was no longer airborne and shut down the motors.

The movement to the southeast is a mystery to me at this stage. The mobile device DAT file (FLY049.DAT) might shed further light on the event.

Mobile device DAT file: How to retrieve a V3.DAT from the tablet

 

[KelloggKid]

SAR104 - First of all a huge heartfelt THANK YOU for looking at the data. I would really like to get to the bottom of this. I have already learned a lot from the incident but am still perplexed as to the root cause. DAT file attached. I might add that we had a big wind storm last night and the drone fell from its perch and is now back home. So if there is data in the drone itself that you could use let me know and I'll try and pull it out.

 

[sar104]

SAR104 - First of all a huge heartfelt THANK YOU for looking at the data. I would really like to get to the bottom of this. I have already learned a lot from the incident but am still perplexed as to the root cause. DAT file attached. I might add that we had a big wind storm last night and the drone fell from its perch and is now back home. So if there is data in the drone itself that you could use let me know and I'll try and pull it out.

This appears to be related to a yaw error and/or an IMU calibration issue:



Note the IMU1 yaw value (green dashed trace). After the initial yaw maneuver at 16 seconds it ends up disagreeing with the compass and IMU0 by 48°. Subsequently, at 46 seconds, it appears that the FC rotates the aircraft counter-clockwise to while holding the IMU1 yaw constant to attempt to resolve the discrepancy. Both IMU z-axis rate gyros and the compass detect the rotation. At 48.5 seconds the FC initiates negative pitch and roll, which is what caused the motion into the tree. The exact reason for that is unclear but likely related to the yaw discrepancy because at 50 seconds the FC reports:

50.309 : 4047 [L-FDI][CTRL]: fault on , horiz_ctrl_fail

The obstacle avoidance system sees the impact coming but the aircraft is not properly under control, and impact occurs at 54.3 seconds:



54.359 : 4250 [L-FDI][CTRL]: fault on , tilt_ctrl_fail
54.378 : 4251 [L-RC]craft ctrl failed!!!
54.438 : 4254 [L-RC]Emergency break. rc vib!
54.558 : 4260 [L-FMU/LED]type:0, normal flash action:
54.558 : 4260 [L-FMU/LED]c0:0,15;c1:0,3;c2:0,13;c3:2,3;c4:0,10;c5:0,3;c6:0,10;c7:0,3
54.857 : 4275 [L-FDI][CTRL]: fault off, tilt_ctrl_fail
55.076 : 4286 [L-FMU/MOTOR]fault on , esc(2) is stall
55.096 : 4287 [L-FMU/MOTOR]fault on , esc(3) is stall
55.894 : 4327 [L-FMU/MOTOR]fault on , esc(1) is stall
56.393 : 4352 [L-RC]craft ctrl failed!!!

The aircraft was manufactured in December 2017, and so presumably is out of warranty, otherwise I'd be calling DJI for a replacement. The on-board data is only readable by DJI.

@BudWalker - any thoughts on this?

 

[KelloggKid]

Wow - what an impressive analysis. The drone was out of warranty and I let DJI Refresh lapse at the end of the first year. The damage appears to be repairable. I get the impression that this is yet another case of the MA's temperamental compass.

 

[BudWalker]

This appears to be related to a yaw error and/or an IMU calibration issue:

View attachment 78246

Note the IMU1 yaw value (green dashed trace). After the initial yaw maneuver at 16 seconds it ends up disagreeing with the compass and IMU0 by 48°. Subsequently, at 46 seconds, it appears that the FC rotates the aircraft counter-clockwise to while holding the IMU1 yaw constant to attempt to resolve the discrepancy. Both IMU z-axis rate gyros and the compass detect the rotation. At 48.5 seconds the FC initiates negative pitch and roll, which is what caused the motion into the tree. The exact reason for that is unclear but likely related to the yaw discrepancy because at 50 seconds the FC reports:

50.309 : 4047 [L-FDI][CTRL]: fault on , horiz_ctrl_fail

The obstacle avoidance system sees the impact coming but the aircraft is not properly under control, and impact occurs at 54.3 seconds:

View attachment 78250

54.359 : 4250 [L-FDI][CTRL]: fault on , tilt_ctrl_fail
54.378 : 4251 [L-RC]craft ctrl failed!!!
54.438 : 4254 [L-RC]Emergency break. rc vib!
54.558 : 4260 [L-FMU/LED]type:0, normal flash action:
54.558 : 4260 [L-FMU/LED]c0:0,15;c1:0,3;c2:0,13;c3:2,3;c4:0,10;c5:0,3;c6:0,10;c7:0,3
54.857 : 4275 [L-FDI][CTRL]: fault off, tilt_ctrl_fail
55.076 : 4286 [L-FMU/MOTOR]fault on , esc(2) is stall
55.096 : 4287 [L-FMU/MOTOR]fault on , esc(3) is stall
55.894 : 4327 [L-FMU/MOTOR]fault on , esc(1) is stall
56.393 : 4352 [L-RC]craft ctrl failed!!!

The aircraft was manufactured in December 2017, and so presumably is out of warranty, otherwise I'd be calling DJI for a replacement. The on-board data is only readable by DJI.

@BudWalker - any thoughts on this?

Just a few observations;
At 37.157 secs the FC decided it had enough info to determine the geoInclination
37.157 : 3387 [L-NS][AHRS] wmm dec: 14.712161
37.157 : 3387 [L-NS][AHRS] wmm inc: 70.665009
The FC started adjusting the Yaw without rotating the AC to reflect the geoDeclination.

I was a little surprised to see this occur a few secs before navHealth reached 4/5.

The roll excursion you referred to didn't actually carry it into the trees. The excursion was terminated at 50.309 when the FC said that horizontal control failed. There after the MA just coasted into the trees. It had almost stopped when the impact came.

The black plot shows the roll excursion with the red and blue plots showing the GPS VelE and VelN. I used the GPS velocities since we're not sure what was going on with the IMU calcs.

@KelloggKid I don't think there was a problem with the compass. Rather, I think it was due to an unfortunate combination of things that caused this. Maybe if the MA were above the trees when the transition to using GPS occurred there would have been enough room for the MA to achieve stability.

 

[sar104]

Just a few observations;
At 37.157 secs the FC decided it had enough info to determine the geoInclination
37.157 : 3387 [L-NS][AHRS] wmm dec: 14.712161
37.157 : 3387 [L-NS][AHRS] wmm inc: 70.665009
The FC started adjusting the Yaw without rotating the AC to reflect the geoDeclination

It started to adjust the IMU0 yaw value, but it held the active IMU1 yaw value constant. I don't understand that strategy.

I was a little surprised to see this occur a few secs before navHealth reached 4/5.

The FC set the declination when the ctrl_level went to 2, and set the home point when the navHealth level went to 4. I'm not sure what the difference between those parameters is.

The roll excursion you referred to didn't actually carry it into the trees. The excursion was terminated at 50.309 when the FC said that horizontal control failed. There after the MA just coasted into the trees. It had almost stopped when the impact came.
View attachment 78275


The black plot shows the roll excursion with the red and blue plots showing the GPS VelE and VelN. I used the GPS velocities since we're not sure what was going on with the IMU calcs.

@KelloggKid I don't think there was a problem with the compass. Rather, I think it was due to an unfortunate combination of things that caused this. Maybe if the MA were above the trees when the transition to using GPS occurred there would have been enough room for the MA to achieve stability.

Yes - it may be that simple, although I'm still puzzled by the IMU1 yaw values.

 

[BudWalker]

It started to adjust the IMU0 yaw value, but it held the active IMU1 yaw value constant. I don't understand that strategy.

I'm thinking it's because IMU(0) was being used and IMU(1) wasn't