Can anyone help me understand this crash?

[Purple Fox]

Hi All

My Dad, my brother and all have Mavic's and have been regularly flying them since early 2017. I have used this website in the early days - thanks to all contributors. I am hoping someone will be able to give me and my Dad an insight into why his Mavic appeared to drop out of the air and crash head on into a sea wall.

Unbelievably it still carried on flying and returned to home but is very badly damaged.

I have included photos, flight log and also a youtube link of the DJI go 4 flight log in the hope that it will help......


We were stood on the Promenade which is elevated over the sandy beach, the tide was well out. This is what happened:
-10:32 am - took off and ascended to about 60 foot
- Flew out over the bay in a straight line
- Then he turned almost through 180 degrees and headed back in towards the sea wall, still not adjusting the height
- It was coming towards him on his left side but he couldn't see it due to the sun so was flying on the screen
- He then pressed RTH
- the next thing the screen went blank (we know know the camera had been smashed off) and he lost video.
- It then came hovering along side us but he had no control, and there was no response from the sticks
- We noticed the camera and gimbal hanging by the ribbon cable
- I grabbed it from underneath and the motors increased to counteract
- eventually it stopped and we turned it off

We then noticed that it was in a really bad shape. The rear rotors were snapped almost completely off, (about 1" remaining, how it flew back to us I will never know). The camera and gimbal had taken an impact at speed and pushed back into the fan area and smashed the main chassis and smashed the fan. The gimbal and camera were ripped off. The front was also heavily scuffed.

We worked out where it had crashed against the wall by using the co-ordinates finding the debris of the props (I think the front arms had bent round on impact and the rear rotors collided, probably then pushing the front arms back out).

It had crashed at a lower height than it took off, IE, into the sea wall to the left of our original position but at no point in the flight log can I see any adjustments on the elevation

1:35 seconds into flight is where I think it crashed looking at the map

He cant remember if he pressed the go home button before or after it crashed (IE before or after the video transmission stopped)

I have uploaded the flight records to Airdata so I can share the flight, I don't know how else to do it (he uses an iPhone and doesnt have iTunes, I am Android).

Here is the link: Airdata UAV - Flight Data Analysis for Drones

Here is the youtube link:

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

CSV also attached (zipped)

If there is anything else I can provide then I will try. I really would be grateful for feedback.



Sea wall here highlighted in red.



For reference, this is a picture I happened to take on mine and shows where we were stood, the crash was obviously out of shot on the right, to our left as we looked out to sea. The sea wall was at least 3m lower than where we were stood but at no point was there a negative elevation.

 

[Xtreme Drone Pilot]

Perhaps, varying air pressure is to blame for poor altitude control? Pressure can sometimes change dramatically in a matter of seconds, I observed barometric altitude change by 500m within a minute in the mountains due to hot pockets of air rising quickly from lower altitudes; something similar can happen on a sea beach where air and waves interact a lot. It's unstable air; expect Mavic to be much more "floaty".

 

[charliesRig]

I was hoping for pictures of damaged Mavic Please

 

[sar104]

It didn't happen quite like you describe, according to the log file. The aircraft was flown at just over 30 mph, in sport mode, into the wall at 95.1 seconds. Its recorded altitude at the time was 15 ft, which it had been holding since climb out after takeoff. RTH occurred at 99.8 seconds, at which point it climbed to its set RTH altitude which appears to have been 30 m.

That sequence of events is also consistent with your screen video, which shows no indication of descent before the crash, just a steady altitude of 4 m, so I'm not clear why you think it "dropped out of the air". At no point is there any evidence of climbing to 60 ft.

 

[Purple Fox]

It didn't happen quite like you describe, according to the log file. The aircraft was flown at just over 30 mph, in sport mode, into the wall at 95.1 seconds. Its recorded altitude at the time was 15 ft, which it had been holding since climb out after takeoff. RTH occurred at 99.8 seconds, at which point it climbed to its set RTH altitude which appears to have been 30 m.

That sequence of events is also consistent with your screen video, which shows no indication of descent before the crash, just a steady altitude of 4 m, so I'm not clear why you think it "dropped out of the air". At no point is there any evidence of climbing to 60 ft.


View attachment 24270

Thanks for taking the time to look at the log and report back. Please don't mis-understand, I am not saying that it wasn't user error, it could well have been but I couldn't see how based on the logs and what I saw.

My mistake - 60 feet should read 6m, sorry. He works in feet, me in meters and the logs have been in both as I look at his iPhone then my wife's which is in meters (the phone in the video is my wifes which is logged into his account as we live at opposite ends of the country).

Your graph shows a sudden drop in altitude, drop in pitch and spike in the roll at the same time - why was this as I can see no control inputs to that effect?

which shows no indication of descent before the crash, just a steady altitude of 4 m, so I'm not clear why you think it "dropped out of the air".
View attachment 24270

That is my point - if it had stayed at 4m, it would have been clear of everything. Where the mavic crashed into the wall it is a good deal lower than where we were stood. If the mavic had stayed at 4m or even 2m, it wouldn't have hit anything at all. It would have to be at -2m for it to hit the wall where it did but yet there is nothing that I can see as to why it dropped so much altitude.

If you see from this photo, we took off on the left which is raised, it crashed into the vertical part of the sea defence to the right - which is a lot lower.



I appreciate your analysis - is there anything further you can deduce given the above explanation?

 

[sar104]

Thanks for taking the time to look at the log and report back. Please don't mis-understand, I am not saying that it wasn't user error, it could well have been but I couldn't see how based on the logs and what I saw.

My mistake - 60 feet should read 6m, sorry. He works in feet, me in meters and the logs have been in both as I look at his iPhone then my wife's which is in meters (the phone in the video is my wifes which is logged into his account as we live at opposite ends of the country).

Your graph shows a sudden drop in altitude, drop in pitch and spike in the roll at the same time - why was this as I can see no control inputs to that effect?

That was the impact.

That is my point - if it had stayed at 4m, it would have been clear of everything. Where the mavic crashed into the wall it is a good deal lower than where we were stood. If the mavic had stayed at 4m or even 2m, it wouldn't have hit anything at all. It would have to be at -2m for it to hit the wall where it did but yet there is nothing that I can see as to why it dropped so much altitude.

If you see from this photo, we took off on the left which is raised, it crashed into the vertical part of the sea defence to the right - which is a lot lower.

View attachment 24306

I appreciate your analysis - is there anything further you can deduce given the above explanation?

The data do not show any drop in altitude but you cannot trust the altimeter to better than a few meters, even though it generally will be more accurate than that. So if it did descend relative to its initial altitude then the FC was no aware of that, and so not attempting to correct for it. This flight simply didn't have enough margin for error, especially if you didn't have a clear view of the aircraft.

 

[Purple Fox]

That was the impact.



The data do not show any drop in altitude but you cannot trust the altimeter to better than a few meters, even though it generally will be more accurate than that. So if it did descend relative to its initial altitude then the FC was no aware of that, and so not attempting to correct for it. This flight simply didn't have enough margin for error, especially if you didn't have a clear view of the aircraft.

I agree that there wasn't enough margin for error - hindsight of course.

Would you really expect a drop of 6m without the Mavic detecting it?

I wonder if being in sport mode was contributory - IE, in normal mode would the forward sensors have increased the altitude automatically.

 

[royster]

According to your description of your after crash mavic, I;m amazed she flew at all, must see pics post crash!

 

[sar104]

I agree that there wasn't enough margin for error - hindsight of course.

Would you really expect a drop of 6m without the Mavic detecting it?

I wonder if being in sport mode was contributory - IE, in normal mode would the forward sensors have increased the altitude automatically.

I would expect the aircraft to detect a 6 m drop, but clearly it didn't. So I'd be inclined to ask whether you are quite sure about those numbers. Heights can be deceptive sometimes. High speeds in sport mode might make altitude measurements less accurate, but that still seems to large an error.

 

[Purple Fox]

According to your description of your after crash mavic, I;m amazed she flew at all, must see pics post crash!

I will get some pics to post up and yes - I too was surprised at the impact it had took and still flew, as well as both rear props being less than half their usual size.

 

[Purple Fox]

I would expect the aircraft to detect a 6 m drop, but clearly it didn't. So I'd be inclined to ask whether you are quite sure about those numbers. Heights can be deceptive sometimes. High speeds in sport mode might make altitude measurements less accurate, but that still seems to large an error.

The heights of the crash site could be questionable? or the heights of the Mavic?

Assuming its the former (as the mavic heights are given in the logs) then I am confident the terrain is as described, I am not local so cannot get further pictures (the one above happens to be on google street view)

Google maps here: Google Maps

Agreed a 6m drop is quite a lot - would there be a case to send the drone to DJI do you think or would they say it should have been higher so was user at fault?

 

[sar104]

The heights of the crash site could be questionable? or the heights of the Mavic?

Assuming its the former (as the mavic heights are given in the logs) then I am confident the terrain is as described, I am not local so cannot get further pictures (the one above happens to be on google street view)

I meant the relative heights of the takeoff point, the wall etc.

Google maps here: Google Maps

Agreed a 6m drop is quite a lot - would there be a case to send the drone to DJI do you think or would they say it should have been higher so was user at fault?

Worth asking DJI, but I doubt they will have much sympathy for 30 mph with a few meters clearance, especially since it's going to be difficult to provide evidence that there was any margin at all.

 

[Strafe1]

It looks like there is an altitude drop in the graph. Altitude was fluctuating by 5-6 feet. In sport mode, the AC will drop a few feet do to drag. You can see it in the graph.

 

[Purple Fox]

All - thanks for the feedback.

 

[f3honda4me]

When you switch to sport mode, you are disabling the VPS which is the most accurate system for keeping the drone at a stable height. Without it you are relying upon the barometer and worse, GPS, which is going to have a margin of error of a few meters. So, this was operating within normal parameters it looks like to me. In addition, sport mode disables the sensors that would detect objects to avoid a collision, such as into a wall.

Not that I think there is much more it would add, but can you grab the DAT file off the mavic? There is more that I might be able to see as that gives me a lot more data than the txt log.

How to retrieve a V3 .DAT File