I had a serous bug when flying my Mavic air today that caused me to lose my drone.  I was flying over a crater today on the big island of Hawaii. I got some really nice footage of the crater and some small lava spots then decided to return. I flew in a straight line and it took about 35% battery to get there.  I started the return back right at 60% left and everything seemed to be going well. When it hit somewhere between 30-40% left the drone requested to return. I was already coming back in a straight line so I cancelled because I wanted to control the return flight myself.  The drone went into a hover and refused to respond.  It hovered for about five minutes and then landed in the crater outside of the semi active crater.  There was no way for me to get it back.  If the drone would have let me control it for even one more minute I would have cleared the crater and I could have walked over to get it.   The drone would not respond got any commands.  

I started with the RC and iPhone and switched to RC and goggles after I started recording.  When the drone refused to respond, I switched back to the RC and iPhone.  It was very frustrating to watch it just hover then land in the crater.  

I have no idea what happened but it appears to be a bug with the return home command being cancelled.  If I lost the drone to by stupidity, then I would be upset but I would understand.  Watching the drone hover and then take my video with it to its grave that I may not be able to try to get again really got me upset.  The small lava spots are expected to be gone very soon and who knows if they will open again.  I was not over the lava spots.

Any ideas on what went wrong would be appreciated.

Hello there. I am sorry to read and to know what happened to your DJI Mavic Air. Since this unfortunate incident happened and you have lost the said drone. I would recommend for you to contact our DJI Support Team at https://www.dji.com/support?site=brandsite&from=nav for further assistance. We have a team that will do there best to help you and give out the best resolution for the said issue. If the said DJI drone is still under the warranty period and it is evaluated as a non-pilot error, the warranty service will be provided. Again, I am sorry for your loss and thank you for reaching out.

I have no idea what happened but it appears to be a bug with the return home command being cancelled.  
Cancelling RTH wouldn't make any difference, it will have been something else.
Post the recorded flight from the incident and we'll see if it has some clues to solve the mystery.

To get to your flight data, go to: https://www.phantomhelp.com/LogViewer/Upload/
... where you'll find instructions to upload the data to that site and it will give a report
For help interpreting the data you can post a link for the report here even if it looks empty or like garbage.
Or just post the .txt file.

Labroides Posted at 6-9 21:06
I have no idea what happened but it appears to be a bug with the return home command being cancelled.  

Cancelling RTH wouldn't make any difference, it will have been something else.

Maybe so but last week I cancelled a return to home command and the drone refused to listen and landed in a bush.

fanse31bd6f9 Posted at 6-9 21:30
Maybe so but last week I cancelled a return to home command and the drone refused to listen and landed in a bush.

Maybe so but last week I cancelled a return to home command and the drone refused to listen and landed in a bush.
And maybe (definitely) that also had nothing to do with cancelling RTH and was caused by something else.
Data from that flight would explain what you are doing wrong.
Cancelling RTH doesn't do anythingnstrange to your flight, I've only been doing it for six years so I do have some idea what I'm talking about.

If you want answers, post flight data.

Labroides Posted at 6-9 21:07
To get to your flight data, go to: https://www.phantomhelp.com/LogViewer/Upload/
... where you'll find instructions to upload the data to that site and it will give a report
For help interpreting the data you can post a link for the report here even if it looks empty or like garbage.

Thanks.  I will work on that when I get home.

Labroides Posted at 6-9 21:38
Maybe so but last week I cancelled a return to home command and the drone refused to listen and landed in a bush.
And maybe (definitely) that also had nothing to do with cancelling RTH and was caused by something else.
Data from that flight would explain what you are doing wrong.

Never say definitely.  It pigeonhole your thinking and often makes you wrong.  I have troubleshoot many turboshaft issues and ran many engineering investigations that lead to thing we didn’t think could happen.

fanse31bd6f9 Posted at 6-9 21:41
Never say definitely.  It pigeonhole your thinking and often makes you wrong.  I have troubleshoot many turboshaft issues and ran many engineering investigations that lead to thing we didn’t think could happen.

Never say definitely.  It pigeonhole your thinking and often makes you wrong.
Cancelling RTH definitely does not make a drone land or do weird, unpredictable things.
These drones are simple devices that do just as they are programmed.
It's the operators that can be hard to work out.

And having confidence in my understanding of the way these things work after 6 years professional flying and analysing the data from hundreds of flight incidents, doesn't "often make me wrong".

Labroides Posted at 6-9 22:07
Never say definitely.  It pigeonhole your thinking and often makes you wrong.
Cancelling RTH definitely does not make a drone land or do weird, unpredictable things.
These drones are simple devices that do just as they are programmed.

Now is not the time to narrow focus.  Unless you are the computer scientist who programmed the device, the hardware engineer that designed the device, and the integrator that make sure the interface controls were followed, I am not sure you have that understanding.  

Time in service isn’t always a good indicator of quality.  I had an engineer that worked for me that had 15 years of experience that was a complete zero and two that proved they were worth their salt in less than one year out of college.  

All we have right now is a pilot interview.  I get those are not the best because pilots are focused on multiple things at one time.  I am working on getting the data and that should help.  We should be data driven and let that rule things out.

fanse31bd6f9 Posted at 6-9 22:20
Now is not the time to narrow focus.  Unless you are the computer scientist who programmed the device, the hardware engineer that designed the device, and the integrator that make sure the interface controls were followed, I am not sure you have that understanding.  

Time in service isn’t always a good indicator of quality.  I had an engineer that worked for me that had 15 years of experience that was a complete zero and two that proved they were worth their salt in less than one year out of college.  

I am not sure you have that understanding.

And I can tell that you definitely don't.
These drones have been around for years now and there have been many flight incidents or all kinds.
Those that look into flight incidents have a pretty good idea of how they work and what can/does go wrong.
Don't go dismissing my expertise just because it's something you have little experience with, and I won't dismiss your expertise in your own field.

I am working on getting the data and that should help.  We should be data driven and let that rule things out.
Looking forward to seeing it and solving more flight mysteries.
If you'd ever seen my analysis of flight incidents, you'd understand that I am very much data driven.

Labroides Posted at 6-9 23:12
I am not sure you have that understanding.  

And I can tell that you definitely don't.

You are misunderstanding.  I am just giving free professional advice you would get if you took a mishap course.  You have made a big assumption on just a pilot interview.  That is an error.  At this point we really know nothing.

When you transition from one flight mode (manual) to another flight mode (automatic pilot) there are a lot of things that can go wrong.  The aircraft itself may be very simple but that does not mean the logic that controls the aircraft based on sensor input and outputs, and the communication between the RC and app is simple.  The number one reason for preflight troubleshooting in aviation is avionics related. This is not a knock on you or your experience, that is just what it is.  

You are correct, I am not qualified to make an assumption if the transition between manual to auto pilot was the problem but I had two incidences in one week where the drone stopped responding when it transitioned between modes.

I can guarantee that when the drone landed itself last week in the bush that it did not take off from that bush.

All I am saying is You just need to keep an open mind when troubleshooting.  If something can happen, given a large enough sample size, it will happen, good, bad, or indifferent.

fanse31bd6f9 Posted at 6-10 00:26
I can guarantee that when the drone landed itself last week in the bush that it did not take off from that bush.

All I am saying is You just need to keep an open mind when troubleshooting.  If something can happen, given a large enough sample size, it will happen, good, bad, or indifferent.

You need to STFU and be gracious. Labroides is one of the most experienced pilots on this forum. He has successfully analysed hundreds of flights to help the owner find out what happened. He does not need 'free professional advice' from someone who has no idea what happened to his aircraft.

This is a user forum, and there is precious little help given by DJI, the vast majority of help is provided by other pilots. They help out free of charge, so they do not need to have to deal with someone who wants to lecture them on procedures. So put up or shut up.

fanse31bd6f9 Posted at 6-10 00:26
I can guarantee that when the drone landed itself last week in the bush that it did not take off from that bush.

All I am saying is You just need to keep an open mind when troubleshooting.  If something can happen, given a large enough sample size, it will happen, good, bad, or indifferent.

You have made a big assumption on just a pilot interview.

The assumption is all on your part.
I didn't assume anything about the cause of the incident and asked (as I always do) to see the data before I say anything about it.
But I did say that your suggestion was definitely not the cause, and I can do that because I understand a whole lot more about it that you do.

All I am saying is You just need to keep an open mind when troubleshooting.  If something can happen, given a large enough sample size, it will happen, good, bad, or indifferent.
And all I'm saying is that you are way out of your field of expertise.
Thanks for your free "professional" advice.
Even free, it's overpriced.

Still looking forward seeing to your data.

Labroides Posted at 6-10 01:47
You have made a big assumption on just a pilot interview.

The assumption is all on your part.

I am sorry that I offended people. That was not my intent and I could have worded it better.  The only thing I was trying to say was to keep an open mind.  I should have just asked why there cannot be an issue when it transitions from manual to autopilot.

I also just wanted to point out that there were two incidences in one week where the drone stopped responding when there was a transition in flight that got cancelled.  I thought that might be useful information.

I should be able to get the data up today.

The bigger question if you already had an previous issue with the drone with RTH when stopped (went in bushes) why did you perform the same maneuver without trying to figure out the cause of the first incident?

videoeditman Posted at 6-10 10:09
The bigger question if you already had an previous issue with the drone with RTH when stopped (went in bushes) why did you perform the same maneuver without trying to figure out the cause of the first incident?

That is a really good question. I should have forced my drone into the conditions to cause it to automatically come home several times to see how it repeated but I didn’t.  The reason I wanted to cancel was because I assumed it would take less battery if it wasn’t relying on auto pilot.  There were no obstacles anywhere nearby since we were in the upper crater of the volcano so it was just a straight path back and I never dropped altitude during flight so I didn’t want to assuming it need to climb before it came back wasting time.  I also had no plans to use the return feature.  

It was a mistake not to troubleshoot that right away.

fanse31bd6f9 Posted at 6-10 10:18
That is a really good question. I should have forced my drone into the conditions to cause it to automatically come home several times to see how it repeated but I didn’t.  The reason I wanted to cancel was because I assumed it would take less battery if it wasn’t relying on auto pilot.  There were no obstacles anywhere nearby since we were in the upper crater of the volcano so it was just a straight path back and I never dropped altitude during flight so I didn’t want to assuming it need to climb before it came back wasting time.  I also had no plans to use the return feature.  

It was a mistake not to troubleshoot that right away.

Hi,

Interesting to know what really happend, if you like post a cloud link to your flightlog on here.
It is for sure not normal that after cancelling RTH the drone cannot be controlled.
But log will tell more....


cheers
JJB

I finally have access to a computer.  Here is the link to the converted long.  I can post the original file too if needed.  

https://www.phantomhelp.com/LogViewer/BFQA78YW6QK06YIMRN7E/

Looks like something happened with communication at 9m 19.3s.  There seems to be a lot of missing data.  

Oh, I just updated my profile and changed my default name from fanse31bd6f9.  

At 1 minute your data stops - 145' from home point why?
#1) Error Message at 9 min 18 seconds Downlink Restored (after 8m 17.4s); High wind velocity. Fly with caution and land in a safe place ASAP.
#2) Error Message at 9 min 19 seconds Extended flight distance detected. Always maintain visual line of sight unless otherwise authorized. (over 5,000' away)
3) Downlink Restored (after 8m 17.4s); High wind velocity. Fly with caution and land in a safe place ASAP.
Time disappeared? 1 min to 9 minutes no data, probably no link, which means your battery was probably much more depleted than you knew!
Also the high wind will suck battery life quickly!
At 9min 19.3 -19.4 seconds the Battery Level says it went from 93% to 19%. (Link re-established)
At 20% it starts to RTH and land.

videoeditman Posted at 6-10 11:40
At 1 minute your data stops - 145' from home point why?
#1) Error Message at 9 min 18 seconds Downlink Restored (after 8m 17.4s); High wind velocity. Fly with caution and land in a safe place ASAP.
#2) Error Message at 9 min 19 seconds Extended flight distance detected. Always maintain visual line of sight unless otherwise authorized. (over 5,000' away)

I think you are on to the problem.  The battery seems to not be operating as it should.  

There were very short gusts of wind where I was standing but there may have been different conditions where the aircraft was.  

I wish I had just switched to sport to give myself a fighting chance before it got too late.  

I have been plagued with compass issues with this aircraft even after degaussing.  Apparently it is a known issue.  

Thanks for the help.  

JJB* Posted at 6-10 10:30
Hi,

Interesting to know what really happend, if you like post a cloud link to your flightlog on here.

Thanks.  You have been very helpful.  

Frito Posted at 6-10 11:01
I finally have access to a computer.  Here is the link to the converted long.  I can post the original file too if needed.  

https://www.phantomhelp.com/LogViewer/BFQA78YW6QK06YIMRN7E/

Hi,

see my chart for your flight.

As said already missing data between 1m01s and 9m18.4 sec

In RTH mode when data is in the log again, drifting slowly backwards (to home heading  = correct)

In RTH mode your MA2 pitched down not that much,wich is little bit weird.
Mayby beacuse of few times the error " Attitude is too large" , wich was not too large. So did this error blocked the max power to the motors ?  i would aks DJI this question.

Many errors in the log

at 1m10s : High wind velocity. Fly with caution and land in a safe place ASAP
Many "not enough force" errors (not seen in the app)

At 10% batt level is started to Autoland, before that in N mode.

Flight started at 98% batt, cell values normal.
At the end cell values normal to the % of the battery, battery oke. (only max power draw to fight the wind)
See the last error " Battery:Overheating During Discharge"

Always check in RTH if the distance to home decreases, if not ; cancel RTH, lower height if possible, into SPORT mode (if N mode does not give enough fwd speed) and apply full stick forward.

cheers
JJB

JJB* Posted at 6-10 12:16
Hi,

see my chart for your flight.

Thanks.  This is an excellent response.  I hope it can help people in the future.  

I tried to get DJI to look at the data but they were having none of it since it was out of warranty.  

Frito Posted at 6-10 11:58
I think you are on to the problem.  The battery seems to not be operating as it should.  

There were very short gusts of wind where I was standing but there may have been different conditions where the aircraft was.  

Sorry for your loss!

videoeditman Posted at 6-10 12:34
Sorry for your loss!

Thanks.  The battery in question was a "new" battery I bought from an authorized dealer.   Must have been old stock.  

I'm a little late to the party as I live on the other side of the world.
Here's how I've reconstructed the incident comparing your description with the recorded flight data.

I started with the RC and iPhone and switched to RC and goggles after I started recording.
This accounts for the missing 8 min 17sec of the flight data.

everything seemed to be going well. When it hit somewhere between 30-40% left the drone requested to return.
We have good data from 9:18.5 where the drone is in RTH and 5020 ft from home and the battery showing 19%.
You had the right stick pushed full forward and the drone was not making headway.
It was being pushed backwards at 0.5-3mph.
This is a clear indication that the drone was pushing against a headwind for the return.

I was already coming back in a straight line so I cancelled because I wanted to control the return flight myself.
You cancelled RTH and resumed control at 9:36.1 (15%)

The drone went into a hover and refused to respond.
The data shows the drone responded normally to your control inputs, but was unable to make headway against a headwind of approx 18 mph, which roughly equates to the top speed of the Air in P-GPS mode.
You had climbed 400 ft down from the volcano's rim to your launch point and had that 400 ft wall behind you blocking the wind to some degree.
But the drone, a mile downwind of where you stood had no protection from that wind.

It hovered for about five minutes and then landed in the crater outside of the semi active crater.
If the drone would have let me control it for even one more minute I would have cleared the crater and I could have walked over to get it.   The drone would not respond to any commands.
You fought against the headwind but were unable to make any progress toward home.
The Air was still responding normally to your control inputs.

I have no idea what happened but it appears to be a bug with the return home command being cancelled.  
At 9:59.7 (10%) your battery reached critical low voltage and the drone commenced autolanding (as it is programmed to).
You continued to push the right stick and as drone descended, it encountered slightly lower windspeed and started to make slow headway.

If I lost the drone to by stupidity, then I would be upset but I would understand.
Perhaps the interpretation of the data helps with understanding how the incident happened.

I would still be very interested to see the data from the previous incident you mentioned.
I'm certain that cancelling RTH did not cause it, but I'd like to find out what did.

Frito Posted at 6-10 14:11
Thanks.  The battery in question was a "new" battery I bought from an authorized dealer.   Must have been old stock.

Nothing in the data indicates any problem with the battery.

Labroides Posted at 6-10 15:23
I'm a little late to the party as I live on the other side of the world.
Here's how I've reconstructed the incident comparing your description with the recorded flight data.

The data is missing where the return to home command was cancelled.  You cannot assume that was when the command was invoked without the data.

Frito Posted at 6-10 17:14
The data is missing where the return to home command was cancelled.  You cannot assume that was when the command was invoked without the data.

The data is missing where the return to home command was cancelled.  You cannot assume that was when the command was invoked without the data.
How about you leave off with your assumptions about me making unfounded assumptions?
It wasn't correct the last few times you made the accusation and it isn't this time either.

I'm not making any assumptions and I have good data upon which I based my analysis.

When data resumes at 9:18.5, it shows your drone in Go Home Mode and continues to show it that way until 9:36.2 when RTH is cancelled and you resumed control.
There is a full data point for every 1/10th of a second during that period and nothing to suggest that some data is missing.

Labroides Posted at 6-10 17:39
The data is missing where the return to home command was cancelled.  You cannot assume that was when the command was invoked without the data.
How about you leave off with your assumptions about me making unfounded assumptions?
It wasn't correct the last few times you made the accusation and it isn't this time either.

I am not sure what your problem is but you need to chill. There is eight minutes of missing data.  You have no idea what happened during that eight minutes.  It is very possible the command was cancelled and the drone reinstated it.  

JJB and Videoeditman made some really good points.  Coupling their comments and what the data looks like to me is there were possible some sensor errors that artificially limited the aircraft.  The aircraft hovered for several minutes in one locations. It is very unlikely there were sustained winds for several minutes.  

I did make a critical error, I did not lock out the aircraft via sport mode.  I do believe there is a critical oversight in the return to home function and with sensor flight.  The aircraft should monitor rate of change of forward speed and throttle position.  If that rate of change is unacceptable, the aircraft should do one of three things or maybe a combination of them.

-Warn the pilot that rate of change is unacceptable and suggest turning on sport mode.
-Flag the pilot that rate of change is unacceptable and raise the limiter.
-Flag the pilot and automatically raise the limiter.  

These are very common practice in aviation and DJI should use them.   Often times there may be more than one limit zone that need to be turned off with the understanding there will be a hardware life reduction.

Not having emergency flight procedures is a critical flaw in the product.  These are not really expensive products but expensive enough there should be emergency procedures logic.

Frito Posted at 6-10 17:50
I am not sure what your problem is but you need to chill. There is eight minutes of missing data.  You have no idea what happened during that eight minutes.  It is very possible the command was cancelled and the drone reinstated it.  

JJB and Videoeditman made some really good points.  Coupling their comments and what the data looks like to me is there were possible some sensor errors that artificially limited the aircraft.  The aircraft hovered for several minutes in one locations. It is very unlikely there were sustained winds for several minutes.  

I am not sure what your problem is but you need to chill.
I don't need to "chill".
I'm not the one who lost their drone because of a lack of basic situational awareness.

There is eight minutes of missing data.  You have no idea what happened during that eight minutes.
It is very possible the command was cancelled and the drone reinstated it.
I don't need to know what happened during those 8 minutes or make guesses about what might have happened (even though they make no difference to the outcome).
I'm only working with the actual data in front of me.  

JJB and Videoeditman made some really good points.  Coupling their comments and what the data looks like to me is there were possible some sensor errors that artificially limited the aircraft.  
It looks to me like you lost your drone because of inexperience and are looking for something else to blame.
But your attempts to assign blame don't stack up.
It wasn't that the drone wouldn't respond to your control inputs.
It wasn't a case of the drone going into autoland when you cancelled RTH.
Your hopes that some kind of drone malfunction can be blamed aren't borne out by the actual data.

The aircraft hovered for several minutes in one locations.
There's no data to show any time where the drone mysteriously hovered for several minutes.
You were pushing the right stick full forward the whole time and the drone was pitching forward at ~20° and achieving a non-zero velocity.
That's not hovering.

It is very unlikely there were sustained winds for several minutes.
Unlikely?  3500 ft up on a mountain top in Hawaii?
You've never heard of trade winds?
The data shows that there was a steady wind from the north east.

I do believe there is a critical oversight in the return to home fiction and with sensor flight.  The aircraft should monitor rate of change of forward speed and throttle position.  If that rate of change is unacceptable, the aircraft should do one of three things or maybe a combination of them.
I believe there's a critical oversight issue with your piloting skill and situational awareness.
And I'm amazed that you continue to attempt to dismiss what's clearly shown in your recorded flight data to avoid taking responsibility for negilgent flying that lost your drone.

Labroides Posted at 6-10 18:17
I am not sure what your problem is but you need to chill.
I don't need to "chill".
I'm not the one who lost their drone because of a lack of basic situational awareness.

I am done with your ridiculous aggressiveness.  I have reported you.

Frito, Labroides means well and is a genius at doing these investigations.....but he comes across a bit coarse. Just read the important information and let the rest go. That is what we do ;-)

Here is a screen shot of your wind map according to airdata.com (might not be that accurate). Higher than the top speed in P mode at the end of the flight.
https://app.airdata.com/share/gPZFLh

Frito Posted at 6-10 19:25
I am done with your ridiculous aggressiveness.  I have reported you.

You don't know when to back off, do you?
Losing a drone can always be stressful.
Losing a drone and not learning from the experience is a double loss.

I am done with your ridiculous aggressiveness.
I'm done with your unique blend of ignorance and arrogance.
I put up with your unfounded assertions that I don't know how to analyse flight data and was making invalid assumptions.
I held off giving you the blunt truth, that your dumb piloting and blissful unawareness of what your drone was doing caused it's loss, and just gave you a dispassionate analysis of the facts revealed by your flight data.
But that wasn't good enough for you and you had to come back with more unfoundded assumptions and desperate guessing to avoid facing the reality of what your incident data shows.
So I put it to you a little more bluntly, still sticking to the facts and that's ridiculous aggressiveness?

I have reported you.
You've also shown anyone that reads this thread what a turkey you are.

And I'd still be very interested to see the data from your previous incident to find out what actually caused it.

GaryDoug Posted at 6-10 19:31
Frito, Labroides means well and is a genius at doing these investigations.....but he comes across a bit coarse. Just read the important information and let the rest go. That is what we do ;-)

Here is a screen shot of your wind map according to airdata.com (might not be that accurate). Higher than the top speed in P mode at the end of the flight.
[Image]

Thanks for the wind map.  The issue I have with wind is the gusts all around we’re about 15-20 seconds long. The drone was stuck for a few to several minutes in that locations. It is possible they were sustained winds in that location though.  

I don’t recall if the app shows wind speed and direction.