Mavic flies into side wall of bridge, uncontrollable whilst 20m away in direct vision. DJI takes no responsibility

[Meta4]

Lol, DJI reimbursed that tree hugging drone as it had compass failure.

Or perhaps you got lucky with DJI.
It wouldn't be the first time they've paid out even though the data shows user errors.
I'm interested in what you mean by "compass failure".
The compass is an extremely reliable component and doesn't just fail.

I don't know what this toxic vibe here is assuming that technology is perfect

You mean that "toxic" vibe of people that believe the recorded flight data rather than the words of someone who has very dubious understanding of how to fly a drone and jumps to blame DJI for their own mistakes??

 

[boblui]

This is what I saw from the flight log DJIFlightRecord_2021-02-14_[11-45-10].txt . The corresponding .DAT file is not provided so the depth of analysis is limited .

The marked time point in the chart below is 1 second before the collision. Full left aileron was applied but the craft kept banked slightly to the right by about 3 degrees. The video shows that the craft moved to the right as well so the banking angle is consistent with the actual movement of the craft.



On the surface this seems unreasonable but let's look at it from the flight controller's perspective - It saw a y-speed of +0.5 m/s. What does it mean ? From what I have seen, the definition of y-speed depends on whether GPS is used :

GPS mode : y-speed is the East component of the horizontal speed. +ve y-speed means the craft has a speed component towards the east

ATTI mode : y-speed is the speed component towards the right of the craft.

Applying these definitions to this case, there are two opposite interpretations :



As shown in the chart, GPS was used at that moment although the signal quality was zero. The craft was still logically in GPS mode. The fused sensor output told the flight controller that the craft was moving to it's left ( light blue direction in the map above ) which is the same as the direction of aileron stick input so things looked OK to the flight controller.

Obviously, the fused sensor output was wrong because the craft was actually moving to the right. I can only attribute the error to poor GPS signals.

So I guess the take-away from this case is don't fly under marginal GPS signals + poor vision positioning conditions which can cause unexpected behaviors of this kind.

 

[Kola]

Expecting technology and the manufacturer to take responsibility for poor understanding and use seldom works out well even if people sometimes get lucky (1 out of 4 in this case).

Expecting the forum to not analyse and trust the actual data also seldom works out well, even less so by judging being thorough as 'toxic'.

 

[Starz]

After reading this entire thread, this old line comes to mind.
There‘s them that has, and them that will.
Not so good things happen when equipment is being used past it’s limits.
Hopefully the op finds the correct equipment for the type of photography and filming he enjoys.

 

[Prismatic]

I don't know what this toxic vibe here is assuming that technology is perfect

No one here ever makes that assumption. On rare occasions, though eventually with certainty, a serious hardware or software fault will cause a drone to crash. But those instances--if you follow this forum--are indeed uncommon, and careful log analysis generally identifies such a fault as the root cause of the incident.

The log files are recorded with robotic impartiality. They don't have imperfect recall, and they have no axe to grind. The facts recorded in your logs in this case are not affected by your wishes, hunches, or beliefs. They tell a very common story--around here at least--of a pilot not realizing he's in way over his head. Been there, done that, and I (hint hint) learned a necessary lesson the hard way.

Now, if you find the facts problematic, you aren't alone, but the facts--stubbornly--don't change to accommodate your outrage.

DJI declined to cater to your fever dreams. So you came here hoping for a more amenable analysis, and now you reject the several expert opinions you've received--just as you rejected DJI's--not because they're wrong, but because you don't like what they reveal.

And then, THEN, you put the cherry on your hubris with that "this toxic vibe here" zinger!

Just ... wow. To quote my late father, "It's a poor workman who blames his tools."

 

[BudWalker]

Just to be clear the claim made by @aquinox that the Mavic moved right when he commanded that it move left is correct. It doesn't look that way when only the latitude, longitude data is considered. @boblui explained this in his post #22. The GPS data was marginal and was telling the FC that the Mavic was moving left when it was not. The FC then commanded that the Mavic move right to compensate.

As for "toxic vibe" that pretty much started in the first sentence of post #2.

@aquinox this type of flying really should be done in ATTI mode where it's not possible for the Mavic to try to use GPS. To do that requires a commonly used modification that changes Sport mode to ATTI mode.

 

[Kola]

Really still do not see a 'toxic vibe', maybe I am just not that sensitive.

Am now more curious to read what drone the OP will get next...

 

[elborak]

As for "toxic vibe" that pretty much started in the first sentence of post #2.

Sorry, I just don't see it. If you want to call it "blunt" or "brusque" or "direct", sure. But toxic? Hardly.

Consider the number of people Meta4 helps on an almost daily basis; would you rather have have him spend time trying to be more tactful but only be able to help half as many?

Some of the best doctors I've ever seen (including the oncologist that saved my wife's life) are blunt, brusque and direct. They're far too busy to pussyfoot around - they have people to help. Different magnitude, but same basic logic here.

 

[Starfighter]

Sorry, I just don't see it. If you want to call it "blunt" or "brusque" or "direct", sure. But toxic? Hardly.

Consider the number of people Meta4 helps on an almost daily basis; would you rather have have him spend time trying to be more tactful but only be able to help half as many?

Some of the best doctors I've ever seen (including the oncologist that saved my wife's life) are blunt, brusque and direct. They're far too busy to pussyfoot around - they have people to help. Different magnitude, but same basic logic here.

I love your analogy to the oncologist here. My sister actually fired one (At Mayo Clinic in Jacksonville for being a cold fish.). It worked out, but some of the most gifted oncologists and are highly gifted technicians, who do their magic magic, save lives, and happen not to be warm and cuddly when dealing with patients. Take the genius!

When analyzing accidents, the accuracy of the analysis trumps a warm and comforting tone.

One thing quickly caught my eye. How do you get the Mavic 2 into atti mode. A mood is mentioned. Atti mode takes practice. Zero experience is asking for trouble when forced into that mode. What’s the mod?

 

[thewishy]

To my mind there are a few things which DJI could improve here
- Making ATTI mode a "Pro" feature which can't easily be accessed on Mavics makes it hard to practice ATTI mode. The closest you'll get is the DJI Flight Simulator, which is better than nothing.
- Warnings that GPS quality is poor / device has switch to ATTI could be a lot clearer (You get annoying beeping for low battery, but not much by way of notification that you're in a degraded flight mode, there should be a warning claxon)
- Some more warnings about flying in built up areas

That said, the Drone did what I'd expect it to do - handle poorly under difficult conditions

 

[colorado kid]

1st time lost...could be a "rookie" mistake.
2nd time lost...some folks learn the hard way.
3rd time lost...there's definitely a pattern developing.
4th time lost...time to look in the mirror.

exactly!

 

[avdk]

My first drone (DJI Air) also had strange behaviour. Flying in open area without any obstacles, GPS signal could vary from 15 satellites to 6 and back to 15. This happened random. The drone was very unreliable to fly this way and I almost lost it when suddenly the GPS signal dropped to 0 (without any obstructions) and the drone started to drift away in the wind. The RTH function didn't work in a situation like that.
I returned the drone to DJI and their conclusion was a defect in the main circuit board. I received a new replacement drone.
Technology can and will fail sometimes.

 

[slup]

...I returned the drone to DJI and their conclusion was a defect in the main circuit board.
Technology can and will fail sometimes.

Yep, that's true ... even if HW failures rarely is the cause for mishaps it happens, all man made thing's can fail.

But in this case the available technology was used outside it's limits & could't work ... can't blame the GPS chip for not picking up satellites when denying it from an obstacle free view of the sky.

 

[aquinox]

This is what I saw from the flight log DJIFlightRecord_2021-02-14_[11-45-10].txt . The corresponding .DAT file is not provided so the depth of analysis is limited .

The marked time point in the chart below is 1 second before the collision. Full left aileron was applied but the craft kept banked slightly to the right by about 3 degrees. The video shows that the craft moved to the right as well so the banking angle is consistent with the actual movement of the craft.

View attachment 125098

On the surface this seems unreasonable but let's look at it from the flight controller's perspective - It saw a y-speed of +0.5 m/s. What does it mean ? From what I have seen, the definition of y-speed depends on whether GPS is used :

GPS mode : y-speed is the East component of the horizontal speed. +ve y-speed means the craft has a speed component towards the east

ATTI mode : y-speed is the speed component towards the right of the craft.

Applying these definitions to this case, there are two opposite interpretations :

View attachment 125099

As shown in the chart, GPS was used at that moment although the signal quality was zero. The craft was still logically in GPS mode. The fused sensor output told the flight controller that the craft was moving to it's left ( light blue direction in the map above ) which is the same as the direction of aileron stick input so things looked OK to the flight controller.

Obviously, the fused sensor output was wrong because the craft was actually moving to the right. I can only attribute the error to poor GPS signals.

So I guess the take-away from this case is don't fly under marginal GPS signals + poor vision positioning conditions which can cause unexpected behaviors of this kind.

That's useful, thanks!
I don't know why the DAT file didn't attach in the first case. Here you go!

 

[aquinox]

Really still do not see a 'toxic vibe', maybe I am just not that sensitive.

Am now more curious to read what drone the OP will get next...

Waiting for the follow up of the Spry+ which hopefully has 1 inch sensor and stabilization. In the meanwhile a mini 2 I guess. I see the dynamic range difference is not so day and night as i thought with m2p.

 

[aquinox]

DJI chose to submit and refund me after all. An additional image given to me on another forum which showed clearly it was aware of my commands but didn't listen was the key

 

[BudWalker]

DJI chose to submit and refund me after all. An additional image given to me on another forum which showed clearly it was aware of my commands but didn't listen was the key

Interesting. Can you provide a link to the other forum posting so that we can take a look?

 

[Kola]

Any reason you cannot share the additional detail or image here?

That way we all learn.

 

[aquinox]

Not always active on the forums :0

Mavic flies into side wall of bridge, uncontrollable whilst 20m a...

Hi,My DJI died whilst flying under a bridge. it moved itself to the right instead of to the front, ignoring my commands while i was 20 meters away, also ignoring my correction (Full throttle) to the left. whilst being very close to the drone. In all other cases it did not pose a problem to fly...

forum.dji.com

 

[BudWalker]

Not always active on the forums :0

Mavic flies into side wall of bridge, uncontrollable whilst 20m a...

Hi,My DJI died whilst flying under a bridge. it moved itself to the right instead of to the front, ignoring my commands while i was 20 meters away, also ignoring my correction (Full throttle) to the left. whilst being very close to the drone. In all other cases it did not pose a problem to fly...

forum.dji.com

I looked at that thread. @JJB44 came to the same conclusion that @boblui ,I and others did. We are all saying the same thing. The Mavic moved in the opposite direction from what you commanded - just as you observed and was shown in your video.

Is this incorrect behavior? Can it be fixed? Should it be fixed? A better way to frame this issue is that the behavior is due to a trade-off made by DJI engineers. Changing the FW to operate better in this situation will cause undesirable behavior in other situations.

I'm surprised that DJI chose to give you a refund. I've analyzed hundreds of incidents, many like this one. This would be the first time DJI chose to accept responsibility for an engineering trade-off. Another Mavic would behave the same in the same set of circumstances.

Allow me to repeat my suggestion that you use ATTI mode for this type of flying.