seems my Mavic 2 Pro needed a submarine mode (or at least be able to float)

[aleksander]

I was attempting to film off a moving sailboat, just off the coast of Spain (about 10nautical miles northwest of Cadiz, where we were offshore heading towards). I had filmed twice already off this same boat the day before (while moving) and the day of (at anchor). As my previous flights were all successful, i was starting to feel like i was get the hang of filming of/from a boat. Boy was i wrong..

As i started the motors up, it wants to take off on me, so i let her go (hand-launching off near the port quarter), and she drifts to the side with the wind (may have been in ATTI mode still, so the wind was taking her away). I was still calm until i notice she seemed to be lowering and was not responding to stick-up, and seconds later, plummets into the ocean. The entire "flight" (if you can call it that) lasted under 10 seconds.. Flight controller said Aircraft Disconnected.. the worst outcome happened :-(

Things are tense (people asking WTF just happened, except in spanish). Within a few moments we turn the boat motor back on, furl away the sails and using the GPS on the controller, and power on back to the last known location where it happened. This was no drill, it was a real "man" (drone) overboard situation, and the depth at this spot was ~25 meters. With no diving gear at hand, short of someone willing to jump into the water and hopefully spotting it underwater, not much hope existed that the aircraft would float to be able to be even recovered, and this lack of hope already began while trying to get back to the spot :-(

Long story short, my Mavic 2 pro could not be recovered and is now "lost at sea" at the bottom near the Gibraltar Strait.. but I am thinking was there perhaps some sort of malfunction as the craft refused to gain altitude on command? Do i have any chance of trying to contact DJI support to ask for a replacement if there's a cause/fault other than my own? I've read about the VPS signal being confused over water as a result lowering, but shouldn't stick-up override that? Replying the flight logs in the App, i see stick input was sent in the final seconds before craft went down so given that I was in sport mode, rate of ascent shouldn't have been a problem either.

Any advice would be really appreciated as I'm wanting to contact DJI support next to see what they'll say.. I'm not a professional, just an amateur here and bought this thing with my own money so shame if this is how this new (~10 month) hobby of mine ends :-/

 

[aleksander]

UPDATE: i've uploaded the .txt file (mobile .dat seems empty and aircraft .dat is unaccessible due to loss of the aircraft)..

What is available from the event is now visible here

To me it looks like the aircraft only updated home point (still acquiring satellites) after 6 seconds, and then got a compass error 9 seconds in (already over 100 foot from the boat), all the while (since about 5 seconds in) trying to endlessly accelerate.. seems it eventually hit >80mph and just flipped over?? Why would it do this, when the winds that day were only a force 2 or 3 (less than 15-20 mph). Not sure why it try to hit that speed.. any ideas?

There's almost 15seconds record of the flight but as mentioned, almost the entire final 5 seconds i was fully pressing stick up to ascend/counteract, as i saw it was slowly losing the little altitude it started with.. :-(

 

[boblui]

The .DAT file attached is the wrong one, the file name should end with "FLY031" and the whole name should be something like 2020-06-27_14-10-28_FLY031.DAT

From the .TXT file, the only anomaly I can see is that the current increased from around 10~ 12A when the craft was holding altitude to as much as 19 A when the craft was descending. Looks like some problem has occurred to the propulsion system. Increase in current did not produce additional thrust.

got a compass error 9 seconds in (already over 100 foot from the boat), all the while (since about 5 seconds in) trying to endlessly accelerate.. seems it eventually hit >80mph and just flipped over?? Why would it do this, when the winds that day were only a force 2 or 3 (less than 15-20 mph).

The compass error made the craft move in the wrong direction when it tried to correct for drift in position. The result was that the more it tried to correct, the bigger the position error became so it accelerated in the attempt to correct the error quickly. If only the compass is faulty, the craft will just run away but in your case, some other problems have developed causing it to descend into the water.

If you can dig out the correct .DAT file, may be some more insight into the cause can be obtained.

 

[BossBob]

One of the problems with flying from a boat, whether you hand launch or launch from the deck, is from the moment you start the motors, the drone will want to hold position. The boat will almost always be moving in one plane or another, drifting or riding the waves. This leads to the drone trying to fly with no input on the sticks, either to maintain its altitude or whatever position it has.

 

[aleksander]

In addition to the log analysis i previously linked to (not sure how long that link is valid for), the more detailed AirData share link is here.

Below are my responses to the replies so far:

The .DAT file attached is the wrong one, the file name should end with "FLY031" and the whole name should be something like 2020-06-27_14-10-28_FLY031.DAT

....

If you can dig out the correct .DAT file, may be some more insight into the cause can be obtained.

I don't have any such DAT files on the mobile device, which end in FLY031 (or any files that have FLY in the name for that matter).

One of the problems with flying from a boat, whether you hand launch or launch from the deck, is from the moment you start the motors, the drone will want to hold position. The boat will almost always be moving in one plane or another, drifting or riding the waves. This leads to the drone trying to fly with no input on the sticks, either to maintain its altitude or whatever position it has.

I would figure this could happen if the boat is in any kind of motion, so i always try to make sure that i am on the leeward side with no obstructions (in other words, on the side where the wind leaves away, not blowing onto/toward the boat). What i saw after hand launching it (it really sounded like it wanted to be let go), was that the drone seemed like it was simply drifting away as if in ATTI mode (but this wasn't the case from the log files) and seemed like it was just moving away at a constant velocity (as it's rather hard to determine it was accelerating when it's just moving away rather than laterally where you can see the distance it's traveling in a perpendicular to your vision plane).

I am wondering if the combination of it attempting to overcompensate endlessly, which resulted in acceleration, PLUS my stick up input just overloaded the abilities of the propulsion and the thrust it was capable of (i saw repeated "Not Enough Force/ESC Errors" in the AirData detailed analysis as well).

At this point any advice on if/how to go about seeing if DJI would issue a replacement? To me it seems that even if there's a compass error, there's unchecked feedback loop when in P-GPS mode and the software should at some point realize something is not right and switch to ATTI mode instead of endlessly trying to correct with no limitations.

 

[peterc4]

Do yu have dji care if not contact dji support you almost certainly will not get a replacement without the drone i would imagine they will offer you 30% of a new one, speaking from very recent experience.

 

[boblui]

I am wondering if the combination of it attempting to overcompensate endlessly, which resulted in acceleration, PLUS my stick up input just overloaded the abilities of the propulsion and the thrust it was capable of (i saw repeated "Not Enough Force/ESC Errors" in the AirData detailed analysis as well).

The message "Not Enough Force/ESC Errors" confirms that there was some problem with the propulsion system. This is the well known "uncommanded descent" problem of the Mini and deformation of prop blades is suspected to be the cause. I am yet to come across the same report for M2. You can apply full forward elevator and full throttle up in sport mode. The overload message may come up but the drone will keep flying.

 

[aleksander]

Do yu have dji care if not contact dji support you almost certainly will not get a replacement without the drone i would imagine they will offer you 30% of a new one, speaking from very recent experience.

Thanks for that experience, I actually have only recently (within last few days) became aware that if i _did_ have DJI Care Refresh that they would required the old drone to be sent back (which wouldn't be possible in this case), so i don't feel as bad having not purchased/registered for this service now.

I am hoping DJI will be of the opinion that it's something that went wrong with the aircraft or the FC malfunctioning and will be covered under warranty (see my last point bout what seems like a bug that this corrective feedback loop has no limitations/failsafe's on it) and will offer a replacement, but i'm of course not holding my breath. Hoping to be pleasantly surprised. If i did wrong something horribly wrong, so be it, but I am not sure how i could have been entirely at fault, other than the very act of engaging in flying it in a risky situation (off a boat, over 25 meters of water), which in and of itself shouldn't invalidate the warranty. The compass warning didn't happen until a few seconds already it was well off the boat, so lacking a crystal ball to know it'll happen after it's been launched, i don't know how i could have prevented any issues :-(

In any case, i've filed a support case and will see what they say.

Is there a thread or some post somewhere anyone knows of for best practices for safe flying from/for filming a boat over water? If there isn't one, i might as well start putting on together as this _has_ been a bit of a learning experience (most unfortunate incidences will be tho).

 

[aleksander]

The message "Not Enough Force/ESC Errors" confirms that there was some problem with the propulsion system. This is the well know "uncommanded descent" problem of the Mini and deformation of prop blades is suspected to be the cause. I am yet to come across the same report for M2.

I've heard of issues with the Mini, i thought it was a design flaw of the mini. Sad if there's other Mavic aircrafts that also can have this issue :-(

You can apply full forward elevator and full throttle up in sport mode. The overload message may come up but the drone will keep flying.

That's the thing, although _I_ was not applying full forward elevator, the craft seems to have been doing this, and while I only applied full throttle up to ascend, it might not have had enough propulsion headroom given it was endlessly accelerating on its own :-(

I'm no aircraft/flight controller engineer (though i have a little electronics/software/computer system engineering experience) but if i had a chance to talk to the DJI engineers who program the FC, my main feedback would be:
1 - don't leave an unchecked feedback loop trying to correct position if it detects it's not helping the situation, maybe switch to ATTI mode (of course this might have OTHER ramifications i am missing, such as trying to combat strong gusts would not recover on its own?), the compass error should have maybe triggered a break in that loop perhaps?
2 - make it at least float a small CO2 canister such as what common recreational boat life-vest use, that inflates automatically on contact with water would be nice (again this would also have hte impact of rasiing the weight) - or does such a product exist?

 

[boblui]

1 - don't leave an unchecked feedback loop trying to correct position if it detects it's not helping the situation, maybe switch to ATTI mode (of course this might have OTHER ramifications i am missing, such as trying to combat strong gusts would not recover on its own?), the compass error should have maybe triggered a break in that loop perhaps?

I have seen some cases in which the compass of M2 was way off or not functioning and the flight controller subsequently commanded the craft to go into ATTI mode. I guess in your case the magnitude of error is not big enough to trigger such change.

The Mini is even smarter. It has build-in intelligence to correct for compass error after it has got the chance to fly around in the GPS grid. I have done a test in which the compass of my Mini was intentionally interfered by a magnet attached to the craft. The orientation of the craft shown on the map was initially way off but after moving the craft around for a while, the Mini was able to correct the error all by itself. Quite amazing.

 

[aleksander]

Impressive indeed, wish my Mavic 2 was this intelligent the-day-of the incident, but going back to this:

The .DAT file attached is the wrong one, the file name should end with "FLY031" and the whole name should be something like 2020-06-27_14-10-28_FLY031.DAT

[...]

If you can dig out the correct .DAT file, may be some more insight into the cause can be obtained.

I am not sure why i am missing the *FLY*.DAT files from my IOS-based DJ Go 4 app.. Per sar104's excellent post which i read fairly thoroughly:

The DJI Fly app deletes the DAT files if and when it syncs flight records with the DJI servers, and so they are often missing from the MCDatFlightRecords folder. The iOS version of the DJI GO 4 app does the same since version 4.3.24. The Android version of DJI GO 4 does not delete the DAT files as of 4.3.32.

I copied all files via iTunes to my Desktop prior to sync'ing (per the info on this thread, tho it seems to relate more to DJI Fly app rather than the DJI Go 4 app i am using), i manually sync'ed and i have the same set of files afterwards (with no *FLY*DAT files anywhere).. Is there anything else that can be done (other than get some scuba gear and go back to Spain), or has this analysis hit a dead-end now without this DAT file?

Also curious, how did you arrive at the "31" in the FLY031 part of what you expected the filename to contain? Nevermind, i see the TXT file converted to CSV has a field named "FLYxxx.DAT index" (except i see this as 39, so guessing 31 was just an example?)

 

[bobec231]

For your next drone.
Check out this link. The drone life preserver is from drone-retriever. com.

 

[aleksander]

Wanted to give folks here with an update to the state of this incident, if nothing else to understand how fast (or slow) they have moved on my situation, and to provide hope and for transparency of their support.

Long story short - after contacting DJI, they agreed this would be cover under warranty and send a replacement.

Here's the full timeline for those interested (note: the date/times i gathered these from are UK-based, so while DJI's US-HQ is in PDT, if i recevied them after midnight they may have actually been from the previous day in california):
- Tues, 7th July - contacted DJI via live-chat, was told to email them, then emailed DJI w/ attachments, no response received
- Wed, 8th July - sensing something was wrong, emailed DJI again this time without any attachments, auto-responder received this time (acknowledging receipt of the case, also mentioning due to covid, repairs are going to take longer)
- Fri, 10 July - first human response, asked for accident date/time/location and details (i already provided most of this in the first email however), also asked for log files.. i responded later overnight
- Mon, 13 July - i sent a follow asking if they received it (worried about attachments causing issues again), half hour later, received a response the nformation has been retrieved and a case has been created, mentioned data validation/analysis will be done in 2-3 business days
- Tues, 14 July - automated feedback/survey email received
- Wed, 15 July - Katy from Data Analysis team informs me "the unfortunate incident...has been confirmed as a warranty case according to our data analysis", asks for contact name/address to send replacement to.. i reply with the details with a request to expedite it since i was trying to lias the replacement with someone flying to the UK (where i'm based out of) end of that week, to hopefully deliver me the replacement (background - i bought it in the US, so they will only ship it to a US address).. but a few hours later, i ask the expediting is no longer needed when i thought of a plan B
- Sat, 18 July - received an automated email saying "Your Mavic 2 Pro has been evaluated by our repair team. Your case No. is... " (this was the day my plan-A contact was supposed to return to the UK but due to personal family reasons had to postpone her trip).
- Tues, 21 July - received a different automated email "After testing your product, DJI has determined to provide you with free repair service.... The repair is expected to be finished within 3 to 6 business days, after which the product will be sent back to you" (this email contains my backup plan's contact name/address)
- Wed, 22 July - received an automated email saying "Your Mavic 2 Pro regarding case No. .... has been processed"
- Thurs, 23 July - received an automated email saying "Your Mavic 2 Pro has been inspected and will be shipped within two business days", then an hour later received a 2nd automated email "Your product for repair regarding case No.....has been shipped via UPS and your tracking number is 1ZE80................", finally hours later, recevied an email from UPS about "Shipment-Details" w/ more info/tracking link, etc..

The shipment type is UPS Ground, and according to the tracking information it was picked up by the sender on the evening of the 23th of July at 5pm, shipped out of Cerritos, CA.. And since my contacts are mainly in NYC, this would normally take 5 more business days or so.. Unfortunately my plan-B contact is now on holiday out of town as of today for a week, so there's still some certainty now that my plan-A original postpone plans are updated to now be planning to fly to the UK by the end of this month, and as such, my plan-B may not receive the package in time to give to plan-A contact to bring to me.. If that's the case, who knows when i'll actually have the replacement back in my hands, but this isn't really a fault of DJI (more of me being impulsive and buying it on a business trip last October in the USA, which means my warranty is only valid in the USA). I am tempted to ask them to change the name of the recipient to plan-A contact, but i managed to at least set the delivery type to collect-from-depot. I just don't have much faith in DJI making this change in time.. We'll see what happens..

Anyway, from first (actual) contacting DJI on the 8th of July until it was shipped on the 23rd of July is just over two calendar weeks (which works out to about 10-12 business days).. Not terrific, and didn't fit into my contacts' schedules, but certainly a good overall outcome, as the incident of the endless-accelerating-then-swimming drone of mine at least wasn't found to be my fault and a no-cost replacement is being sent!

I definitely learned a thing or two from this incident and going to look into some kind of retrieval system (like those small yellow water-deployment lines intended for saving lost fishing-rods.. the getterback i think?). Also going to be a LOT more careful about the compass and not rush into a fly without doing more safety checks/considerations.

 

[Crouching Leopard]

Wanted to give folks here with an update to the state of this incident, if nothing else to understand how fast (or slow) they have moved on my situation, and to provide hope and for transparency of their support.

Long story short - after contacting DJI, they agreed this would be cover under warranty and send a replacement.

Here's the full timeline for those interested (note: the date/times i gathered these from are UK-based, so while DJI's US-HQ is in PDT, if i recevied them after midnight they may have actually been from the previous day in california):
- Tues, 7th July - contacted DJI via live-chat, was told to email them, then emailed DJI w/ attachments, no response received
- Wed, 8th July - sensing something was wrong, emailed DJI again this time without any attachments, auto-responder received this time (acknowledging receipt of the case, also mentioning due to covid, repairs are going to take longer)
- Fri, 10 July - first human response, asked for accident date/time/location and details (i already provided most of this in the first email however), also asked for log files.. i responded later overnight
- Mon, 13 July - i sent a follow asking if they received it (worried about attachments causing issues again), half hour later, received a response the nformation has been retrieved and a case has been created, mentioned data validation/analysis will be done in 2-3 business days
- Tues, 14 July - automated feedback/survey email received
- Wed, 15 July - Katy from Data Analysis team informs me "the unfortunate incident...has been confirmed as a warranty case according to our data analysis", asks for contact name/address to send replacement to.. i reply with the details with a request to expedite it since i was trying to lias the replacement with someone flying to the UK (where i'm based out of) end of that week, to hopefully deliver me the replacement (background - i bought it in the US, so they will only ship it to a US address).. but a few hours later, i ask the expediting is no longer needed when i thought of a plan B
- Sat, 18 July - received an automated email saying "Your Mavic 2 Pro has been evaluated by our repair team. Your case No. is... " (this was the day my plan-A contact was supposed to return to the UK but due to personal family reasons had to postpone her trip).
- Tues, 21 July - received a different automated email "After testing your product, DJI has determined to provide you with free repair service.... The repair is expected to be finished within 3 to 6 business days, after which the product will be sent back to you" (this email contains my backup plan's contact name/address)
- Wed, 22 July - received an automated email saying "Your Mavic 2 Pro regarding case No. .... has been processed"
- Thurs, 23 July - received an automated email saying "Your Mavic 2 Pro has been inspected and will be shipped within two business days", then an hour later received a 2nd automated email "Your product for repair regarding case No.....has been shipped via UPS and your tracking number is 1ZE80................", finally hours later, recevied an email from UPS about "Shipment-Details" w/ more info/tracking link, etc..

The shipment type is UPS Ground, and according to the tracking information it was picked up by the sender on the evening of the 23th of July at 5pm, shipped out of Cerritos, CA.. And since my contacts are mainly in NYC, this would normally take 5 more business days or so.. Unfortunately my plan-B contact is now on holiday out of town as of today for a week, so there's still some certainty now that my plan-A original postpone plans are updated to now be planning to fly to the UK by the end of this month, and as such, my plan-B may not receive the package in time to give to plan-A contact to bring to me.. If that's the case, who knows when i'll actually have the replacement back in my hands, but this isn't really a fault of DJI (more of me being impulsive and buying it on a business trip last October in the USA, which means my warranty is only valid in the USA). I am tempted to ask them to change the name of the recipient to plan-A contact, but i managed to at least set the delivery type to collect-from-depot. I just don't have much faith in DJI making this change in time.. We'll see what happens..

Anyway, from first (actual) contacting DJI on the 8th of July until it was shipped on the 23rd of July is just over two calendar weeks (which works out to about 10-12 business days).. Not terrific, and didn't fit into my contacts' schedules, but certainly a good overall outcome, as the incident of the endless-accelerating-then-swimming drone of mine at least wasn't found to be my fault and a no-cost replacement is being sent!

I definitely learned a thing or two from this incident and going to look into some kind of retrieval system (like those small yellow water-deployment lines intended for saving lost fishing-rods.. the getterback i think?). Also going to be a LOT more careful about the compass and not rush into a fly without doing more safety checks/considerations.

Sorry to hear about your drone. I know the "lost child feeling". Great outcome. DJI are pretty quick once they get the drone and the data files off your mobile device. Most of the time consuming is in the file analysis and shipping. I have used the service twice. Their Australia support centre is only 40 Ks away so I have delivered myself. Really friendly. Assessment and file analysis took about 4 days. Return of drone was as fast as I can (legally) drive, otherwise, UPS is about as fast as it gets.

My suggestion would be if regularly launching from boats, get a removable flotation device. I never launch from boats as, for me, the are just sick-making machines?

 

[icecoldgin]

Good to know that your crash was treated as a warranty issue thus a replacement for a lost drone.

Looking at your timeline, it may be long but here it is definitely longer. My Mini crashed 18 November and up to now, I still do not have my replacement Mini. Got a call from the dealer yesterday and they are still waiting for feedback from DJI. Looks like I will be getting this in about 2 month's time yet. Oh well, que horror, que barbaridad