The other day I had my first ever experience with a DJI fly away. After inspecting the Flight Record and importing into Airdata, I am still baffled as to why my Mavic 2 Pro had a full blown panic attack and flew away on me. I was flying for a snowboarding project in the backcountry, with no cell service but decent GPS signal.

PLEASE - if anyone has experienced something like this and is good at deciphering Airdata, I would greatly appreciate the help here. My hope is to go back in the summer and find the aircraft to retrieve the media from this project on this day... I would like to import the .csv into Google earth/maps, save maps for offline, hike in, and look at my location in relation to the last known point of the aircraft. Can provide tons of data from flight, etc.

Thank you people!!

[backcolor=rgba(249, 249, 249, 0.3)]https://app.airdata.com/share/bcGQYr

Also would like to know if @DJI can do something about this. Why did this happen? All battery voltage was good, fully charged, very light winds. Completely went out on its own and was not responding to any of my controls and I tried to bring it back to take off location.

My hope is to go back in the summer and find the aircraft to retrieve the media from this project on this day...
The last location recorded for the drone was 43.19183   -110.89283.
But this isn't where it ended up as at the time, it was 45 ft lower than where you launched and flying at 50 mph on a curved path.
It could gone quite some distance before it crashed into something.

I would like to import the .csv into Google earth/maps, save maps for  offline, hike in, and look at my location in relation to the last known  point of the aircraft. Can provide tons of data from flight, etc.
Here's what your flight data looks like:  https://www.phantomhelp.com/logviewer/Q4P3NYXPLGJTLNHF7D67/
There's a link there to give you a KML file to show the flight (the part with GPS data) on Google Earth.

Also would like to know if @DJI can do something about this. Why did this happen? All battery voltage was good, fully charged, very light winds. Completely went out on its own and was not responding to any of my controls and I tried to bring it back to take off location.
There were many issues with this flight.

1.  You launched with a battery that had been sitting and discharging.
The battery % indicator gives a false reading in that situation and your battery was almost empty.
It's effectively chasing its tail just trying to sit in one place.

After only 25 seconds, under the load and climbing, the battery cell voltages had almost dropped to the critical low voltage level of 3.3 volts.

2.  You didn't wait for GPS before launching.
Although you might have had the green GPS- Ready to Fly on the top of the screen your drone had no GPS location data until 39 seconds into the flight. (I've seen this a few times lately).
A check would be the distance shown on the screen.
Until there is a home point, it will show as N/A and a distance can't show until the drone has recorded its home point.

And it wasn't until 45.7 secs that the GPS location data was good enough for the drone to record a home point which will have been some distance from where you launched.

3.  - This is why the incident happened
39 seconds into the flight, just as the drone starts to get GPS location data, it takes off and accelerates to higher than normal speed on a smooth curved flight path.
This is the classic evidence of a yaw error at work.
It happens because you powered up the drone at a spot where some steel close to the drone deflected the compass so it was reading wrong.
The IMU gyro sensor is initialised from the compass at power-up.
But if the compass is reading wrongly, the gyro will get an incorrect heading.
When the drone moves away from the steel or magnetic influence that affected the compass, it will read correctly, but the gyro won't and there will be conflicting direction data from these sensors.

When the drone gets location data, the flight controller has a problem because the position data shows movement in one direction, but the gyro shows the drone has moved in a different direction.
The drone tries to correct which puts it further off course and it enters a never-ending correction phanse that causes the drone to zoom away at high speed.
In this state the drone cannot hover in place and is uncontrollable.
It's effectively chasing its tail just trying to stay in one place.

Hi tgraustin, I am sorry to hear about the accident that happened to your drone. You can try to check the last location of the aircraft in the accident flight record on the app.
Please contact our support to start a ticket, we will help to process data analysis if the drone is under warranty period (within 12 months old), if it's a warranty case, we will compensate the drone according to After-Sales Policy.
Here’s the link to contact us: http://www.dji.com/support

DJI Susan Posted at 3-9 20:02
Hi tgraustin, I am sorry to hear about the accident that happened to your drone. You can try to check the last location of the aircraft in the accident flight record on the app.
Please contact our support to start a ticket, we will help to process data analysis if the drone is under warranty period (within 12 months old), if it's a warranty case, we will compensate the drone according to After-Sales Policy.
Here’s the link to contact us: http://www.dji.com/support

You can try to check the last location of the aircraft in the accident flight record on the app.
That would be completely pointless as explained in the previous post.

Labroides Posted at 3-9 19:24
My hope is to go back in the summer and find the aircraft to retrieve the media from this project on this day...
The last location recorded for the drone was 43.19183   -110.89283.
But this isn't where it ended up as at the time, it was 45 ft lower than where you launched and flying at 50 mph on a curved path.

"The battery % indicator gives a false reading in that situation and your battery was almost empty."

That is really interesting. Is that true for all Mavic drones or just the M2? Can you provide some details or a link to a discussion about it?

GaryDoug Posted at 3-9 21:55
"The battery % indicator gives a false reading in that situation and your battery was almost empty."

That is really interesting. Is that true for all Mavic drones or just the M2? Can you provide some details or a link to a discussion about it?

It's true for DJI "smart" lipo batteries.
Flight data from many, many flights confirms it.
Compare the % indication with the cell voltage.

Hi,

Had a look at your flight too.
Last part your MP2 decended in a  sort of 'linear' way....with vertical speed from 10 m/s to almost 1 at the end.

Landing / crash position not far from last recorded latLon in the log, see my chart.Hope you will find your drone

Green mark = HP set, before that line starts where your drone received GPS data in flight (after 39s), beginning of the flight no GPS  flying in OPTI mode.
If you like put a cloud-link to your DAT file for this flight, more info (probably about compass and power error). FLY060.DAT on your device.

cheers
JJB

GaryDoug Posted at 3-9 21:55
"The battery % indicator gives a false reading in that situation and your battery was almost empty."

That is really interesting. Is that true for all Mavic drones or just the M2? Can you provide some details or a link to a discussion about it?

Hi Gary,

Both charts at 97% battery!

This flight and a random picked flight by me for a MavicPro2, bad and normal cell values.
Shows the importance to check cell values before and after take-off, this a normal routine for me!

cheers
JJB

Labroides Posted at 3-9 19:24
My hope is to go back in the summer and find the aircraft to retrieve the media from this project on this day...
The last location recorded for the drone was 43.19183   -110.89283.
But this isn't where it ended up as at the time, it was 45 ft lower than where you launched and flying at 50 mph on a curved path.

Hi Labroides,

Thank you for your input, and for taking the time to analyze my flight data. It is important to know that the only time I had control of the aircraft was during take-off, from initial ground launch to about 10 ft off the ground. After that, the aircraft fully flew on its own with no response from my RC.

The last location recorded for the drone was 43.19183   -110.89283.
But this isn't where it ended up as at the time, it was 45 ft lower than where you launched and flying at 50 mph on a curved path.
It could gone quite some distance before it crashed into something.
- Should I be looking in the forest roughly 45ft below the last known GPS coordinates? I'm confused as to how the aircraft could have travelled even further beyond the last known point...


There's a link there to give you a KML file to show the flight (the part with GPS data) on Google Earth.
- Copy. I had been trying to import the .csv files, not the KML file. Since this is a remote backcountry location, I really need to be able to download high resolution mapping for offline use, so I can see my location in relation to the flight path.
- My question for you is, if GPS was not 100%, then how much of this data can be trusted? Will it be off by hundreds of feet, if not more? The other crew members who were on the shoot had visual line of sight during the fly away and where we initially searched was not close to where this flight path is showing. How much of an error could be in play here?


You launched with a battery that had been sitting and discharging
- How do you know this? The batteries that were used had been used for 3 days in a row, with 3 days of full use and recharging to 100% each night before the shoot day. There were no "inconsistent firmware" warnings or anything. I believe my batteries are good, so I'm curious how you can come to this conclusion that they were giving a false reading. The battery had not been sitting and discharging.


You didn't wait for GPS before launching
- I recalibrate the GPS whenever I take off from a new location. Always. Even if I move 300 feet I still recalibrate for this exact reason. I successfully calibrated the GPS before this flight and had the green "Ready to Fly"... So, why would it tell me GPS calibration succeeded when really what you're seeing is I had no GPS data until 39 seconds into the flight? That is very strange you've seen this a few times lately...!


3.  - This is why the incident happened
- Classic evidence of yaw error at work. I powered up the aircraft mid slope on a remote mountain side far from steel structures. Have you heard of rock interfering with the compass? Only thing around me was rock, trees, and a lot of snow. It makes sense that if the aircraft eventually regains correct compass while in the air, it will be "confused" while communicating with the RC that is still on the ground within the interference it was causing from take off.. I'm really just confused as to WHY this all happened, even after following a correct pre-flight routine that I've done for years.


ANY input on where to actually look in a few months would be greatly appreciated. Can the last known point GPS coordinates be trusted, or can I use them as a trustworthy baseline to help locate the fallen aircraft?  


Appreciate it!

GaryDoug Posted at 3-9 21:55
"The battery % indicator gives a false reading in that situation and your battery was almost empty."

That is really interesting. Is that true for all Mavic drones or just the M2? Can you provide some details or a link to a discussion about it?

Seriously.. Quite scary!

JJB* Posted at 3-10 01:55
Hi,

Had a look at your flight too.

Hey JJB, thank you. How did you create the green 3D flight path in Earth? And you were able to come up with a new estimated crash spot how?

tgraustin Posted at 3-10 09:12
Hey JJB, thank you. How did you create the green 3D flight path in Earth? And you were able to come up with a new estimated crash spot how?

Hi,

Green 3D flight path is created by my own analysis program.
Just see this chart with all the important flight data.

See the move line shortly after GPS data is recieved, craft starts to move bc of compass/yaw errors.

Looking at your RC input, it is with some marging possible to guess your takeoff spot. After receving GPS data, HP setting and ofcourse baro height takeoff GE ground elevation data : possible to estimate/calculated landing/crash spot,
assuming that the descend in the last part did not stop ofcourse. if it did stop than it your craft would move on until it met higher ground, fly path in the going down in height curve.

Ground level heigt difference : HP (GE values: 2002m), last recorded postion(1979m) = 23 meter lower. Drone baro fly height HP = 64mtr, height last position = -14mtr.
calc: 2002+64-78=1988, so at last position craft about 10 meters above ground. Flypath further is uphil, so your drone cannot be far from my calculated spot.Takeoff spot about 40 meter higher than HP postition, so craft was flying higher above ground than 10 meters.
As said flying away futher in the last curved direction, ground is going uphill. if we know the actual takeoff spot than its possible to calc if drone would fly over that hill or not.
But as it was going down last part, my guess is has landed earlier.

About your battery ; at start your 4 cells above 4 volts!  so if you had check the cell values in the app at takeoff you would not have been alarmed.But shortly in flight all cell dropped quickly to not normal values, see my other post #11 comparing your 97% cell values with a good battery.
Was it extremely cold that day ? A good battery is so important to a drone, if your battery is from 2018....batt don`t have eternal life, so my advice is to change old batts for new ones after few years.

cheers
JJB

tgraustin Posted at 3-10 09:12
Hey JJB, thank you. How did you create the green 3D flight path in Earth? And you were able to come up with a new estimated crash spot how?

Load kml. to GEarth.

When open, right click on path then on properties.

When properties box opens click on extend path to ground  

tgraustin Posted at 3-10 09:09
Hi Labroides,

Thank you for your input, and for taking the time to analyze my flight data. It is important to know that the only time I had control of the aircraft was during take-off, from initial ground launch to about 10 ft off the ground. After that, the aircraft fully flew on its own with no response from my RC.

Should I be looking in the forest roughly 45ft below the last known GPS coordinates?
I have no idea because I cannot tell how high above terrain (+ tree height) the drone was at the last data point.
The height is relative to the launch point which wasn't recorded by the drone.
Show where you launched from and I might be able to make an estimate.

I'm confused as to how the aircraft could have travelled even further beyond the last known point...
Because when the data stopped, your drone was still flying at 50 mph.
If the drone had plenty of clearance above trees and terrain, it could have kept flying quite a distance after signal was lost.

- My question for you is, if GPS was not 100%, then how much of this data can be trusted?
How much of an error could be in play here?
There's no reason to distrust the GPS data, it should be accurate to within a couple of metres.
But we don't know how much further the drone flew after you lost signal.

I believe my batteries are good, so I'm curious how you can come to this conclusion that they were giving a false reading. The battery had not been sitting and discharging.
Your recorded flight data tells a different story.
The battery had partially discharged or it was a very sick battery.
Look at the cell voltages here:  https://www.phantomhelp.com/logviewer/Q4P3NYXPLGJTLNHF7D67/

After only 25 seconds, under the load and climbing, the battery cell  voltages had almost dropped to the critical low voltage level of 3.3  volts.
It should have still had ~4V in each cell.

- I recalibrate the GPS whenever I take off from a new location. Always. Even if I move 300 feet I still recalibrate for this exact reason. I successfully calibrated the GPS before this flight and had the green "Ready to Fly"... So, why would it tell me GPS calibration succeeded when really what you're seeing is I had no GPS data until 39 seconds into the flight?
You have never calibrated your GPS .. there is no way to calibrate the GPS and it never needs calibrating.
You have been recalibrating the compass which would have no effect on the GPS and was completely unnecessary.

That is very strange you've seen this a few times lately...!
It was a false notification of GPS status, but I mentioned how you could confirm that you really had GPS by checking the indicated distance from the homepoint - no homepoint = no distance.

I powered up the aircraft mid slope on a remote mountain side far from steel structures. Have you heard of rock interfering with the compass?
I'm really just confused as to WHY this all happened, even after following a correct pre-flight routine that I've done for years.
Underlying geology would probably have no effect unless you were on a mountain of haematite.
I wasn't there so I don't know what caused the compass deflection.
But the data clearly shows what happened.
Given your confusion about compass/GPS and completely unnecessary compass recalibration, I wouldn't count on your pre-flight routine to achieve anything.

ANY input on where to actually look in a few months would be greatly appreciated.
Point out where you launched from

Can the last known point GPS coordinates be trusted, or can I use them as a trustworthy baseline to help locate the fallen aircraft?
They can be trusted, but as explained, they might be nowhere near where the drone ended up.

Another angle: