DJI Mini 3 Pro | Uncontrolled sinking leads to ditch into ocean despite countermeasures

[matzeulrich]

Hey there,

some days ago my DJI Mini 3 Pro ditched into the ocean despite I was pushing the up-stick.
Because I saw some whales nearby the beach I was very excited and ran to get my drone, quickly started it and took off from a place I used for take off on the days before. At this moment I was very excited about the whales, so I forgot to check if the drone had received GPS Signal.
After take off I flew the drone in sport-mode with full forward movement towards the whales above the ocean in an estimated height of about 3-5m. I did this kind of flying the days before without any problem, too.
Suddenly I received a red warning on the remote that the drone was flying in ATTI-Mode. I was surprised by that, because I never experienced this behavior before. At this moment I also realized the drone was loosing altitude without me pulling the stick to sink. As a countermeasure I pushed the left stick to raise the altitude but nothing happened. I thought that maybe through the steep forward angle the drone could not gain altitude so I stopped the forward movement of the drone by releasing the forward stick. After this I pushed the left stick again to gain height. But again nothing happened. I released the left stick again to give a new impulse to climb, but again nothing happened. this also was the moment I saw the drone ditching into the water. After some few seconds the signal to the drone was lost.
Sadly I could not retrieve the drone from the ocean so all I have is the flight log from the controller. I already uploaded it to Phantomhelp.com but I can't figure it out what happened wrong in this accident.

The whole flight only lasted about 40s, and everything went really quick.

Here is the link to my flight-log analysis.

DJI Flight Log Viewer - PhantomHelp.com

In this folder I uploaded the original flightlog .txt file and the exported .csv file.

2024-01-29_DJI Mini 3 Pro Ocean-Ditch

Shared with Dropbox

www.dropbox.com

I didn't file a fly-away case at DJI yet, because I first want to understand with your help what happened and what I may have done wrong, and if there may is a case of technical malfunction of the drone itself.
So I really appreciate your help. If you need any further information just tell me and I will supply them asap.

 

[Dronebike]

I fortunately saw this video before I flew too close to water. I’m staying at 20 ft to hopefully avoid it.

 

[Phantomrain.org]

You took off before you got GPS confirmation. That for me is Pilot Error
Even at the crash you only had 8 satelites.

Phantomrain.org
Gear to fly in the Rain. Land on the Water.

 

[Yorkshire_Pud]

At this moment I also realized the drone was loosing altitude without me pulling the stick to sink

Was this from watching the drone of looking at the height displayed on the screen?

 

[slup]

I fortunately saw this video before I flew too close to water. I’m staying at 20 ft to hopefully avoid it.

Its just Pilot error , you took off before you got GPS confirmation.
Even at the crash you only had 8 satelites.

Neither water or flying in ATTI mode affects the height hold...

...if there may is a case of technical malfunction of the drone itself.

Hi there & welcome to the forum

Yeah... this doesn't look normal, the drone doesn't behave as expected.

Below is a part of the log that demonstrate this... you were without horizontal positional hold (ATTI) but that doesn't effect the height hold.

-The dashed black graph is the throttle input (left stick forward/backward, neutral stick have value 1024, and full ascending 1684).

-The dashed green is the elevator input (right stick forward/backward... neutral stick have value 1024, and full forward flight 1684).

-The blue graph shows the barometric height above the takeoff location.

-The red graph is the vertical speed (here called zSpeed)... a hover have 0mph, a descent have positive values & a ascent negative values.

-The magenta graph is the pitch angle... negative = nose down for forward flight, 0 flat for a hover & positive = nose up for backwards flight.

(Click on the chart to make it larger)

Where I've placed the chart marker is a good spot to look at... From the blue we see that the barometric sensor is registering a descent... the black on the other hand (your throttle stick) shows that you command full throttle for a ascent. The vertical speed, the red graph, agrees with the blue barometric graph... it also says a descent (value is positive). So the drone totally disregards your stick input.

A couple seconds later you let off from the throttle, but maintains the green elevator for full speed forward... the drone still descends & the vertical descending speed actually increases (shown both by the blue & red graph).

During the whole time the pitch angle (magenta) have been maintained on max allowed for Sport mode... 40 degrees. We can't unfortunately verify the heading speed the drone reaches in a reliable way as you were in ATTI mode.

The uncommanded descent behavior suddenly stops when you release the elevator stick (green)... but the drone still doesn't listen to your full stick command for a ascent (black) according to the red vertical speed (speed is very near 0mph)... the blue barometric sensor though, says that the height increases.

The first that springs to mind is... can this be a "uncommanded descent" case coming from flattened prop blades? But in those cases the drone never could reach full commanded negative pitch... so not similar in this case. And we haven't the heading speed so we can't verify anything from that.

Can it be front props that are flattened & doesn't generate enough thrust..? That would explain why the drone can pitch forward with ease, but in the process not enough total thrust is generated to keep the drone airborne. But would think that the height hold would be of a higher priority then achieving a full specified pitch angle?

In the end of the chart we also see that the vertical speed (red) says hover but the barometric height (blue) indicate a ascent... this instead indicate sensor or IMU problems.

I'm pretty sure of one thing though... this isn't a pure pilot error, think you should make contact with DJI about this.

 

[Yorkshire_Pud]

In some cases the mini 3 pro has a readable DAT.
I can't remember if the newer versions of the App encrypt the DAT but the App version in this log seems to be 1.7.4 and I have a readable Mini 3 Pro DAT from app version 1.9.4 .

Where was the flight flown, the log suggests Mexico but that's only in the last 3 lines

It might be useful if the DAT was posted.

 

[Meta4]

I fortunately saw this video before I flew too close to water.

The video guy is confused and contributing to the myth about downward sensors getting "confused" over water.
Any confusion would relate to horizontal position holding, but not vertical.
If he'd looked into this flight data, he might have found the actual cause of his drone landing over water.
But he gives no indication that he looked into his flight data at all.

What he says about how the sensors measure height is not correct.
Neither is his hypothesis about rebooting the drone and hovering.
This guy isn't someone to trust for technical drone information.

 

[NoBozos]

Sorry for the loss of the drone.

Pilot error, specifically impulsivity, is the dangerous attitude causing contributing to the crash.

Understand the feeling of excitement to get such a stunning shot.

 

[Yorkshire_Pud]

Sorry for the loss of the drone.

Pilot error, specifically impulsivity, is the dangerous attitude causing the crash.

Understand the feeling of excitement to get such a stunning shot.

Did you read what Slup wrote?

 

[matzeulrich]

Hey,
thanks for the welcome and the detailed analysis of the flight @slup!
This was very helpful to see the technical details in these graphs and getting your interpretation and opinion on this case!

I can understand that it seems like a fully pilot error because of flying without GPS lock, and I'm fully aware of this fault on my side, but I really don't get why the drone continued descending even I though I pushed the throttle (left stick) to 100%. I want to understand this topic because in my considerations the drone really shouldn't have reacted like this.

Was this from watching the drone of looking at the height displayed on the screen?

I think I noticed the descent through the drones camera image. Maybe also through the height indication on the RC, but I definitely wasn't watching the drone itself at this moment.

Where was the flight flown, the log suggests Mexico but that's only in the last 3 lines

The flight took place in Mexico, that's fully right. The last recorded, and only recorded position fits to the crash position of the drone.
Here is a Google-Maps Screenshot with the estimated route of the drone:

It might be useful if the DAT was posted.

I would love to post the DAT, but sadly automatic synchronization to DJI was turned on and I connected the controller to wifi, so the folder where these files are supposed to be is empty. Normally I would do some data recovering, but until now I didn't find any way to mount the filesystem of the DJI Remote on my Computer. So this is no option right now to get the DAT file.


I will post this case in the DJI Forum, too, maybe there are some more people with great analyzing skills. I will post the link to this thread here, and vice versa.
Here is the link to the thread in the DJI forum:
Uncontrolled sinking leads to ditch into ocean d...

Thank you so far! I will get in touch with DJI about this case. Does anyone have hint for this? Directly file a "Fly-Away" case or better talk to the customer support first?

 

[NoBozos]

Did you read what Slup wrote?

Yes.

Theses issues MAY have been discovered on inspection, checking inputs and responses at the home point.

My goal here isn’t to put salt in the wound. It’s to learn and recognize dangerous attitudes. The pilot displayed impulsivity. To what extent that caused the accident, is open to debate. But it’s at least possible that these errors could have been detected before the crash. The errors occurred in <40 seconds.

It is also possible that the crash had nothing to do with the pilot‘s impulsivity. But it is to be recognized and avoided.

OP, good luck resolving it. Hope you’re back flying soon.

 

[StevenBrodsky]

Hey,
thanks for the welcome and the detailed analysis of the flight @slup!
This was very helpful to see the technical details in these graphs and getting your interpretation and opinion on this case!

I can understand that it seems like a fully pilot error because of flying without GPS lock, and I'm fully aware of this fault on my side, but I really don't get why the drone continued descending even I though I pushed the throttle (left stick) to 100%. I want to understand this topic because in my considerations the drone really shouldn't have reacted like this.



I think I noticed the descent through the drones camera image. Maybe also through the height indication on the RC, but I definitely wasn't watching the drone itself at this moment.


The flight took place in Mexico, that's fully right. The last recorded, and only recorded position fits to the crash position of the drone.
Here is a Google-Maps Screenshot with the estimated route of the drone:
View attachment 172347


I would love to post the DAT, but sadly automatic synchronization to DJI was turned on and I connected the controller to wifi, so the folder where these files are supposed to be is empty. Normally I would do some data recovering, but until now I didn't find any way to mount the filesystem of the DJI Remote on my Computer. So this is no option right now to get the DAT file.


I will post this case in the DJI Forum, too, maybe there are some more people with great analyzing skills. I will post the link to this thread here, and vice versa.

Thank you so far! I will get in touch with DJI about this case. Does anyone have hint for this? Directly file a "Fly-Away" case or better talk to the customer support first?

File a fly away case. You may have jumped the gun when you flew off to film whales, but it seems apparent that the drone itself is at fault. It didn't respond to your ascent stick commands. The water had nothing to do with it.

 

[StevenBrodsky]

Yes.

Theses issues MAY have been discovered on inspection, checking inputs and responses at the home point.

My goal here isn’t to put salt in the wound. It’s to learn and recognize dangerous attitudes. The pilot displayed impulsivity. To what extent that caused the accident, is open to debate. But it’s at least possible that these errors could have been detected before the crash. The errors occurred in <40 seconds.

its also possible that the crash had nothing to do with the pilot‘s impulsivity. But it is to be recognized and avoided.

to OP, good luck resolving it. Hope you’re back flying soon.

I see what you're saying, but DJI isn't going to take into account the pilot's impulsivity. If the drone didn't ascend on command, the drone didn't ascend. DJI has no idea the condition of the props. They only have the log data to go by. I would expect DJI to fine in favor of the OP. They will look at the logs only, not the drone's physical condition or the pilots rush to film.

 

[slup]

I can understand that it seems like a fully pilot error because of flying without GPS lock...

Think it's well established from your attached log that this wasn't a pilot error... & flying without a proper GPS lock doesn't cause uncommanded descents.

As said earlier... it's 2 likely causes to this, either the drone couldn't produce enough thrust to keep it airborne or respond to your throttle input... or it was sensor/IMU malfunctions.

It's impossible to go deeper than this as we are lacking a reliable heading speed as your drone was in ATTI mode & that we miss a DAT log which otherwise could provide us with more clues through the motor data.

...Does anyone have hint for this? Directly file a "Fly-Away" case or better talk to the customer support first?

If the drone still is within warranty I would start out with that... meaning that you shouldn't pay anything, which is the case if you go with the Flyaway insurance. If the warranty have expired, your only option will be Flyaway as you can't send in the drone which usually is needed for a ordinary DJI Care case..

 

[NoBozos]

I see what you're saying, but DJI isn't going to take into account the pilot's impulsivity. If the drone didn't ascend on command, the drone didn't ascend. DJI has no idea the condition of the props. They only have the log data to go by. I would expect DJI to fine in favor of the OP. They will look at the logs only, not the drone's physical condition or the pilots rush to film.

Yeah.

For the record I hope and expect DJI to take care of this. That’s the warranty side. It’s separate from the safety issue, even though they’re related. As for warranty, feels like a “fly away” to me.

And the OP and I are the same in the sense that — believe me — I too am susceptible to these same whims. We all are. So my goal here isn’t an ad hominem on the OP. It’s to have an honest conversation about what could have possibly avoided the situation.

We can only control what we can control. We know that things fail. So had the pilot undergone the usual checks prior to departing the home point (unrecorded here) the pilot would have controlled what he could. Like a walk around doesn’t guarantee no mechanical failures. It might help you detect an issue.

I’m here to learn.

 

[matzeulrich]

Thank you guys for your further participation in this case!

I really know, that I made a mistake by flying without waiting for the satellites and also for not doing the right pre-flight checks. I'm totally aware I maybe could have prevent this crash by doing these steps before flying over the sea.

For me the most interesting part in this case is the difference between my inputs on the RC and the actual behavior of the drone.

I now filed a fly-away report at DJI and will keep you updated on this in the next days. Let's see. Thank you all for your help and condolences

 

[Meta4]

I really know, that I made a mistake by flying without waiting for the satellites and also for not doing the right pre-flight checks. I'm totally aware I maybe could have prevent this crash by doing these steps before flying over the sea.

You have been mislead by some here.
Neither of those would have made any difference at all.

 

[NoBozos]

I don’t know for certain a quick check would have caught the issue. Maybe. Maybe not. It did occur within 40 seconds. So, at least maybe.

And it could have stopped responding during the check at the home point and flown on its own to the exact same spot and sank. Anything’s possible.

To my way of thinking it was the only possible thing you could have done that MIGHT have detected the issue.

Others have different opinions. And that’s cool. That’s cool.

Good luck. Please update with what you learn from DJI.

 

[dajomu]

Regarding rushing to get the great shot:

I recall listening to a podcast interview with a commercial drone pilot who shot for the Discovery Chanel and National Geographic. I can't recall his name. But I do recall him saying (effectively): "we always bring multiple drones into the field, and if we don't lose one or two, it probably means we aren't trying hard enough to get the great shots."

I don't personally follow that advice, and I don't advocate impulsive or irresponsible flying. I'm an old pilot who has been flying commercially for years, including for film companies. I've never lost a drone, and haven't crashed for years. But, then again, I'm not flying for the Discovery Chanel either. ;-)

 

[matzeulrich]

Today I received an answer from DJI regarding my flyaway case:

"Flight time: 2024-01-27 22:13:04 GMT+8

MC SN: XXXXX FLY(90) BAT SN: XXXXX



1. The aircraft worked under Sport mode after took off.

2. Flight Time T=00:40, Relative Height H=5.0m, The GPS signal was weak, affected its positioning accuracy, the aircraft came into contact with the water.

3. Please wait until the aircraft gained enough GPS signal to operate the aircraft.



As a result, it was determined that the incident was due to a non-manufacturing defect. We therefore regret to inform you that this case is not eligible for a free replacement under the warranty."


For me this isn't the answer I hoped for. I thought there would be some information about the cause that led to the ditching.
If it is really just my piloting mistake I would love the hear the specific reason to learn from this for the next time, but as far as I don't understand why the drone didn't react to my RC commands I can't believe that it's just a piloting error.
What do you guys think?

I answered the DJI mail with the specific questions about the drones missing reactions to my RC commands and why this could have happened.
Let's see