Fly-away and crash at 30mph...

[Fracapouille]

And ...without GPS to provide location data for each image, how did you think you could do photogrammetry?

You don't need positions or attitude info to process pictures with photogrammetric software such as pix4d or Photoscan (or even camera calibration - but it's more debatable).
Even if you record these values, they are just guidelines for the software to help it understand the model and increase processing time, but anyway they are not used as measurments in the processing, all those values are re-calculated. This is what makes photogrammetry with drones efficient, as the tools to measure those values are far from being accurate enough for traditionnal photogrammetric processing. It is compensated mainly by increasing picture overlap and by algorithms based on tie points matching between pictures to calculate all the parameters for each picture (position, attitude and even camera internals).

You can do very accurate photogrammetry from any pictures set, as long as you have strong overlap and a homogenuous ground control points network. I have modeled an indoor miniature park using different camera models (even smartphone), 5000+ pictures (~10 days calculation), none of them geotagged, and just a scale and orientation factor (and a few manual tie points)... I have managed to deliver a survey from the available pictures during the flight that lead to the crash, even if overlap was, for my taste, a bit too low.

 

[sar104]

Hi, all!


Finally, after sending the drone to DJI on the 17th of august, I had a reply from DJI... yesterday.
So, more than 2 monthes later, here is this answer:

“SN: 07JDE1C00101HC incident date: 09/08/2017 14:42 (GMT+8)

1. Due to the weak GPS signal, the aircraft was working in optical mode after taking off, and finally changed to Atti mode;
2. The aircraft cannot hover and position in Atti mode, and the avoidance system cannot work;
3. At t=13:54，h=38m，The aircraft crashed and caused the attitude imbalanced;

Conclusion: pilot fault, non warranty is suggested.”

tl;dr : In ATTI mode, you're f**ked up.

I'm disappointed by the lack of concern and feedback from DJI about this case. It's now a 1000€ (~1200$) bill to get it replaced through RMA.
This whole experience leaves me with quite a bitter taste...

My personnal conclusion is that there is still a very long way to get those birds as reliable as an eBee, for exemple, which I can still trust after 400+ flights and 5 years of duty.
For professionnal purposes, especially in photogrammetry - i.e. you don't fly where you would like to, you fly where you have to - I think this is not a proper device, except if you consider those as consumables, and if you budget to buy one replacement every 6 or 12 monthes. An if you wear an helmet, just in case...

Anyway, feel free to share your feedback here, I would appreciate to know how you would react in my place...

Thanks!

Thanks for the update. Given the circumstances of the flight (no GPS) and the fact that the flight data indictate that the aircraft responded appropriately to all flight commands, I'm not surprised by their response. The mystery of the final direction of travel is not going to be apparent to them because of the complete lack of GPS data, so this will look like a flight that was launched in ATTI mode and then flown into an obstacle.

Obviously that's not very satisfying but I don't think it reflects on the reliability of the Mavic in general. You chose to launch in an environment that, if not well outside its intended operating envelope, is certainly at the most challenging end of the spectrum. At the very least, you flew it at a distance that while within line of sight, made it hard to visually determine orientation. It was a somewhat risky endeavor.

 

[With The Birds]

Thanks for the update. Given the circumstances of the flight (no GPS) and the fact that the flight data indictate that the aircraft responded appropriately to all flight commands, I'm not surprised by their response. The mystery of the final direction of travel is not going to be apparent to them because of the complete lack of GPS data, so this will look like a flight that was launched in ATTI mode and then flown into an obstacle.

Obviously that's not very satisfying but I don't think it reflects on the reliability of the Mavic in general. You chose to launch in an environment that, if not well outside its intended operating envelope, is certainly at the most challenging end of the spectrum. At the very least, you flew it at a distance that while within line of sight, made it hard to visually determine orientation. It was a somewhat risky endeavor.

If your analysis is right it wouldn’t matter if the AC crashed right under his nose the outcome with DJI would be the same. How far the AC was away and how risky you think the flight was is irrelevant.

Btw- he wasn’t flying a Mavic, perhaps your response was meant for another thread.

 

[sar104]

Probably true regarding the outcome with DJI. My comment regarding the distance was simply in relation to how hard it can be to fly in ATTI if you are not close enough to see direction and orientation.

I had just been on the Mavic forum, hence mentioning the wrong aircraft. Correct thread though.

 

[AirMan05446]

View attachment 86703

A lot of that will buff right out.

 

[BudWalker]

A lot of that will buff right out.

Towards the end of building our house I would kid my wife. I'd tell her that I'd discovered the drain pipes under the slab were all wrong and that some load bearing walls were omitted. But, no problem, the painters would fix all that.

 

[Fracapouille]

Thanks all of you for the feedback.

Just to clarify, I did expect DJI to avoid the issue regarding the payback, and I'm not disappointed there.

But I'm more concerned that they didn't investigate a bit more and just answered: "Ok you're in ATTI mode, you shouldn't have, if anything happens it's your fault".

I mean, if the drone is not supposed to be operated in ATTI mode, just lock it in GPS and the story ends. That being said, I'm a land surveyor, I'm very aware of the GPS reliability. You can't just say rely on the GPS and everything will be fine, GPS won't work in many situations, and even when it does work, there's nothing for granted in terms of accuracy.

Anyway, did I miss something, and do DJI specify somewhere that from the moment you operate in ATTI mode, then you're out of warranty bounds?
I didn't use external software or install devices, I just used it in normal conditions (including security)...

If a car manufacturer has a model with both automatic and paddle-shift gearbox, and if gears change erratically in paddle-shift mode, could it say to its customer "it is meant to be used automatic, if anything happens in paddle-shift mode, it's none of our business"?

It seems to me that DJI should ensure its drones reliablity in any mode, provided it has been used in correct "state of the art" conditions.

 

[Meta4]

But I'm more concerned that they didn't investigate a bit more and just answered: "Ok you're in ATTI mode, you shouldn't have, if anything happens it's your fault".

I mean, if the drone is not supposed to be operated in ATTI mode, just lock it in GPS and the story ends. That being said, I'm a land surveyor, I'm very aware of the GPS reliability. You can't just say rely on the GPS and everything will be fine, GPS won't work in many situations, and even when it does work, there's nothing for granted in terms of accuracy.

Anyway, did I miss something, and do DJI specify somewhere that from the moment you operate in ATTI mode, then you're out of warranty bounds?
I didn't use external software or install devices, I just used it in normal conditions (including security)...

It seems to me that DJI should ensure its drones reliablity in any mode, provided it has been used in correct "state of the art" conditions.

Yes ... you did miss something. Here's what DJI says:
DJI warrants that each DJI product that you purchase will be free from material and workmanship defects under normal use during the warranty period.
DJI don't say anything at all about using atti and warranty.
The thing is that your incident was not due to material and workmanship defects under normal use.

It will be the same with your car.
The maker will guarantee it against any faults or malfunction, but they won't cover you against driving into a brick wall on an icy road.

If you were flying in atti mode and the Phantom crashed due to a malfunction caused by a DJI problem, they would cover it.
But fly in atti mode (or any other mode) and crash into a rock wall when the Phantom's systems are working perfectly, and DJI will say that's your problem.

Flying in atti mode is quite safe in many situations, but using atti mode close to a cliff edge and at a distance, is very risky and even for an experienced pilot it could be asking for trouble.
DJI didn't knock back your claim because you flew in atti mode.
They rejected it because the crash was not due to anything DJI could have had any control over.

 

[Fracapouille]

The thing is that your incident was not due to material and workmanship defects under normal use.

If you were flying in atti mode and the Phantom crashed due to a malfunction caused by a DJI problem, they would cover it.

They rejected it because the crash was not due to anything DJI could have had any control over.

How can you be so affirmative about that? The analysis in this post tends to lead to the conclusion that I didn't have full control over the device, and that's what I experienced as pilot.
Of course, everything is more simple if you consider the pilot as a dumb, or a liar, or a reckless fool.
Everything is great then, DJI devices are perfectly reliable, and no one ever complains, as those who actually are complaining fit systematically in the 3 categories above.

Please don't take me for a fool, re-read the thread, and try to be a bit less Manichean, would you?

It seems that the device didn't respond correctly to my controls, and I wish I had an explanation for that, even if it was -partially or totally - my fault. I'm OK to assume whatever I've done wrong, I've already made mistakes (sometimes costly), and learned from that. Prove me I'm wrong, and I'll be totally satisfied with that.

But don't tell me not to fly nearby cliffs, as it is my job to do so, and the reason I bought a DJI trusting its reliability. Don't tell someone whose brakes may have been faulty: "you shouldn't have been on a mountain road".
I'm not using UAVs for entertainment, I'm using them for specific, and very often dangerous missions, in harsh environments (remember robot's work: "Dull, Dirty, Dangerous"). I need to trust as much as possible the equipment, because I have many things to manage and monitor when on the field for a survey. The last thing I need to hear, is someone telling me: "you desserved that, you where flying near a cliff in ATTI mode". Of course I was, that's what drones should be used for in the professional field...

 

[Meta4]

Please don't take me for a fool, re-read the thread, and try to be a bit less Manichean, would you?

And you can read what I posted again. I didn't say anything about you.
I told you why DJI did not accept your claim.

It seems that the device didn't respond correctly to my controls, and I wish I had an explanation for that, even if it was -partially or totally - my fault. I'm OK to assume whatever I've done wrong, I've already made mistakes (sometimes costly), and learned from that. Prove me I'm wrong, and I'll be totally satisfied with that.

Flying in atti mode is like driving on ice with no brakes.
Your drone can't stop, it can't hold position and drifts with wind and inertia.
It will appear to not respond properly to your controls - that's what happens in atti mode.
Flying in atti mode close to a cliff is like driving on ice with no brakes close to a rock wall.
It's asking for trouble and if you do it at a great distance and with little experience, even more so.

But don't tell me not to fly nearby cliffs, as it is my job to do so, and the reason I bought a DJI trusting its reliability. Don't tell someone whose brakes may have been faulty: "you shouldn't have been on a mountain road".
I'm not using UAVs for entertainment, I'm using them for specific, and very often dangerous missions, in harsh environments (remember robot's work: "Dull, Dirty, Dangerous"). I need to trust as much as possible the equipment, because I have many things to manage and monitor when on the field for a survey.

Don't fly close to cliffs without GPS position holding because it's asking for trouble.
Your brakes weren't faulty - in atti there are no brakes. That's a limitation of atti mode and you chose to fly that way.

The last thing I need to hear, is someone telling me: "you deserved that, you where flying near a cliff in ATTI mode". Of course I was, that's what drones should be used for in the professional field...

Sorry but it's the most important thing you need to hear.
You don't drive your car over a cliff because you know that it has no anti-gravity feature.
You need to learn that if you fly in atti mode, your drone has no ability to stop or hold position and can easily be crashed if there are obstacles nearby.
Understand the limitations of your Phantom and work within those limitations. Don't try to make it do things it can't just because you'd like it to.

 

[sar104]

How can you be so affirmative about that? The analysis in this post tends to lead to the conclusion that I didn't have full control over the device, and that's what I experienced as pilot.
Of course, everything is more simple if you consider the pilot as a dumb, or a liar, or a reckless fool.
Everything is great then, DJI devices are perfectly reliable, and no one ever complains, as those who actually are complaining fit systematically in the 3 categories above.

Please don't take me for a fool, re-read the thread, and try to be a bit less Manichean, would you?

It seems that the device didn't respond correctly to my controls, and I wish I had an explanation for that, even if it was -partially or totally - my fault. I'm OK to assume whatever I've done wrong, I've already made mistakes (sometimes costly), and learned from that. Prove me I'm wrong, and I'll be totally satisfied with that.

But don't tell me not to fly nearby cliffs, as it is my job to do so, and the reason I bought a DJI trusting its reliability. Don't tell someone whose brakes may have been faulty: "you shouldn't have been on a mountain road".
I'm not using UAVs for entertainment, I'm using them for specific, and very often dangerous missions, in harsh environments (remember robot's work: "Dull, Dirty, Dangerous"). I need to trust as much as possible the equipment, because I have many things to manage and monitor when on the field for a survey. The last thing I need to hear, is someone telling me: "you desserved that, you where flying near a cliff in ATTI mode". Of course I was, that's what drones should be used for in the professional field...

Unfortunately @Meta4 is correct. There is no clear evidence that the aircraft malfunctioned, but it is clear that you were flying in ATTI mode with limited situational awareness. We don't have good data on the aircraft's flight path because it had no GPS, but we can tell from the pitch, roll and yaw data that it was responding appropriately to stick inputs. It appears most likely that you became disoriented in respect of the aircraft orientation and applied the wrong stick corrections. But, either way, all DJI will see is that the aircraft was responding to commands and that it flew into a cliff in ATTI mode. On what grounds would you expect them to cover that under warranty?

It doesn't matter whether you are using this equipment professionally or not if your expectations of its capabilities plus your capabilities are unrealistic. Just because you are using it in a hazardous environment is not going to give it magical powers to correct for pilot errors. No one is saying that you deserved to crash - just that you knowingly ran a risky assignment and should have been aware that losing control was a possible outcome. DJI is not insuring you against accidents - they are warrantying the aircraft against defects.

 

[Fracapouille]

It will appear to not respond properly to your controls - that's what happens in atti mode.

OK.
Basically, you're telling me that in ATTI mode the UAV will do whatever it wants (I'm not talking about hovering with wind and inertia, that is normal attitude), and that you can't fly a P4P without GPS.
My bad, I should have realized that DJI products were fully dependant on GPS, and unable to fly without.
Maybe I should have chosen a more appropriate UAV.

Would any of you have some advice about UAVs that the pilot himself can control, without relying only on GPS (or even better, without GPS), at a medium price range?
If possible with sensors to avoid collisions that would be activated when you need them the most (i.e. when flying the device in difficult conditions, without GPS of course)?
I need something that can fly nearby bridges, cliffs, buildings, indoors, appropriate for various photogrammetry missions...

Flying in atti mode close to a cliff is like driving on ice with no brakes close to a rock wall.

To sum up a car with brakes, a responding steering wheel, and bumpers... and snow tires as on option.


Thanks by advance for your suggestions!

 

[msinger]

Would any of you have some advice about UAVs that the pilot himself can control, without relying only on GPS (or even better, without GPS), at a medium price range?

The Intel Falcon 8+ is pretty impressive. I saw a demo of it at InterDrone this year. It's able to remember and avoid previously seen obstacles (no matter the aircraft's orientation). I don't think you're going to be able to find tech like that for DJI consumer drone money though.

 

[Meta4]

Basically, you're telling me that in ATTI mode the UAV will do whatever it wants (I'm not talking about hovering with wind and inertia, that is normal attitude), and that you can't fly a P4P without GPS.
My bad, I should have realized that DJI products were fully dependant on GPS, and unable to fly without.
Maybe I should have chosen a more appropriate UAV.

No .. I'm telling you that in atti mode, a Phantom can't do whatever you want.
I've gone over this already but you don't seem to have taken anything in.
Without GPS, the Phantom can't tell where it is and the obstacle sensors can't operate.
The Phantom can't stop and it can't hold position.
That's just the way it is.
With GPS it will do all you want and more.
Solution: realise the limitations and don't expect it to do things it can't.
The problem is your (mis)understanding of what it can and can't do - not a design fault.
I doubt you'll find a more suitable drone without spending 10 x as much and maybe not even then.

Would any of you have some advice about UAVs that the pilot himself can control, without relying only on GPS (or even better, without GPS), at a medium price range?
If possible with sensors to avoid collisions that would be activated when you need them the most (i.e. when flying the device in difficult conditions, without GPS of course)?
I need something that can fly nearby bridges, cliffs, buildings, indoors, appropriate for various photogrammetry missions.

You want something that doesn't exist.

 

[Fracapouille]

We don't have good data on the aircraft's flight path because it had no GPS, but we can tell from the pitch, roll and yaw data that it was responding appropriately to stick inputs. It appears most likely that you became disoriented in respect of the aircraft orientation and applied the wrong stick corrections. But, either way, all DJI will see is that the aircraft was responding to commands and that it flew into a cliff in ATTI mode. On what grounds would you expect them to cover that under warranty?

Thanks for your balanced reply. I admit I could have been disoriented. But the drone can't?
I mean, if I understand correctly, both gyro and compass didn't record any U-turn from the drone. Either both are wrong, and the drone made this U-turn, and crashed NW, or I'm a liar and the drone went to crash SE. The crash impact is on the right side of the drone. I, and some workers that where there when it crashed, can testify about that place it has crashed, but that's all I've got...

So no, I didn't expect them to cover that. What I want most is an explanation, and believe me, I totally accept being eventually guilty for what happened. I understand that it leads to a dead-end for the moment. But hey, you guys listened to it and spent some time analyzing the situation, and I'm grateful for that. You seem to have done much more than DJI, even if I understand perfectly that these guys can't investigate each individual case as much as you did.
But that's also a form of denial that there can't be any problem in ATTI mode... Regardless of my case, how can you ask for a legitimate responsability of the company if you crashed having done nothing wrong in ATTI mode then? How would you feel, except in an unfair situation, when you can't even ask the manufacturer to look a bit further, and when the definitive answer is "Pilot fault"? Not even a small space for a reasonable doubt? No, just the "Pilot fault"? ATTI + crash = Pilot fault, break it up, nothing to see.

Just because you are using it in a hazardous environment is not going to give it magical powers to correct for pilot errors. No one is saying that you deserved to crash - just that you knowingly ran a risky assignment and should have been aware that losing control was a possible outcome.

What if there were no errors? If I had been an expert pilot, would you have the same reaction? Yes I took risks, and yes, I was aware that a wind blow, a bird flying too close, a rock falling, or mainly, some wrong instructions on my side, could have lead to a crash. It already has. I have seen it going a bit erratic, hovering, nothing wrong there, that can be compensated by the pilot. It requires to be smooth, and I certainly need to improve there. But I would never had expected a drone going to its death to the exact opposite direction I was imposing on the controls... Never.

 

[Fracapouille]

No .. I'm telling you that in atti mode, a Phantom can't do whatever you want.
I've gone over this already but you don't seem to have taken anything in.
Without GPS, the Phantom can't tell where it is and the obstacle sensors can't operate.
The Phantom can't stop and it can't hold position.
That's just the way it is.

OK, OK, let's calm things down.
It is totally accepted on my side that the drone won't know where it is in ATTI mode. If the wind blows, the drone moves, and its inertia makes it slower to rest in a position.
The sensors can't operate, OK. This sensor thing is something I think is a bad conception decision, but maybe there are some strong reasons behind that.

I had my drone hovering slowly, no wind at all, normal attitude. The front of the drone was facing me (pictures prove that). I pushed the control fully right, and the drone kept accelerating to its left (my right) to reach 80mph.
How could ATTI mode alone explain that the drone went this fast to the opposite direction I was imposing, that is my question.

 

[sar104]

Thanks for your balanced reply. I admit I could have been disoriented. But the drone can't?
I mean, if I understand correctly, both gyro and compass didn't record any U-turn from the drone. Either both are wrong, and the drone made this U-turn, and crashed NW, or I'm a liar and the drone went to crash SE. The crash impact is on the right side of the drone. I, and some workers that where there when it crashed, can testify about that place it has crashed, but that's all I've got...

So no, I didn't expect them to cover that. What I want most is an explanation, and believe me, I totally accept being eventually guilty for what happened. I understand that it leads to a dead-end for the moment. But hey, you guys listened to it and spent some time analyzing the situation, and I'm grateful for that. You seem to have done much more than DJI, even if I understand perfectly that these guys can't investigate each individual case as much as you did.
But that's also a form of denial that there can't be any problem in ATTI mode... Regardless of my case, how can you ask for a legitimate responsability of the company if you crashed having done nothing wrong in ATTI mode then? How would you feel, except in an unfair situation, when you can't even ask the manufacturer to look a bit further, and when the definitive answer is "Pilot fault"? Not even a small space for a reasonable doubt? No, just the "Pilot fault"? ATTI + crash = Pilot fault, break it up, nothing to see.




What if there were no errors? If I had been an expert pilot, would you have the same reaction? Yes I took risks, and yes, I was aware that a wind blow, a bird flying too close, a rock falling, or mainly, some wrong instructions on my side, could have lead to a crash. It already has. I have seen it going a bit erratic, hovering, nothing wrong there, that can be compensated by the pilot. It requires to be smooth, and I certainly need to improve there. But I would never had expected a drone going to its death to the exact opposite direction I was imposing on the controls... Never.

The uncertainty is in the location, not the direction of travel at impact. If you recall my analysis, without GPS location data all I could do was integrate the IMU velocity data which are, themselves, integrated acceleration data. Those resulting, double-integrated data, disagreed with your description, and were therefore incorrect, but there should be no expectation that they will be correct given cummulative sensor bias errors.

However, at any time during the flight we do have instantaneous heading, pitch and roll data, and those are fully consistent with the stick inputs that are also recorded. At the end of the flight, the aircraft was pointing to the northeast and flying to the right (southeast), and accelerating because you were applying right aileron. It was not ignoring your stick inputs - you were simply not aware of its orientation even though that would have been displayed on your device.

So, based on the data analysis, the only confusion is over location but, since that did not cause the crash and since you were flying, deliberately, in ATTI, that is not an aircraft fault. To reiterate, there is no evidence in the logs that the aircraft ever did anything contrary to stick inputs.

 

[sar104]

OK, OK, let's calm things down.
It is totally accepted on my side that the drone won't know where it is in ATTI mode. If the wind blows, the drone moves, and its inertia makes it slower to rest in a position.
The sensors can't operate, OK. This sensor thing is something I think is a bad conception decision, but maybe there are some strong reasons behind that.

I had my drone hovering slowly, no wind at all, normal attitude. The front of the drone was facing me (pictures prove that). I pushed the control fully right, and the drone kept accelerating to its left (my right) to reach 80mph.
How could ATTI mode alone explain that the drone went this fast to the opposite direction I was imposing, that is my question.

If you have photos that show that the aircraft was facing SW, i.e. towards you, just before the crash, then I'd like to see them. It was not travelling to its left at the end - the roll data unambiguously show that it was moving to its right.

 

[Fracapouille]

The Intel Falcon 8+ is pretty impressive. I saw a demo of it at InterDrone this year. It's able to remember and avoid previously seen obstacles (no matter the aircraft's orientation). I don't think you're going to be able to find tech like that for DJI consumer drone money though.

I often have a stand next to Trimble at various events dedicated to land surveyors, and they're promoting the same model branded as Trimble. I've been discussing with them for a while. Quite an impressive tool, but you're right regarding the price, that's an expensive toy.
senseFly's albris seemed quite well equiped as well, but very expensive too, and with quite a bad customer feedback.

In terms of hardware, I would say there's everything required on a P4P (sensors, camera specs,...), that's why I thought it would suit my needs. But it seems I would need something that can behave better (respond?) out of GPS range. Maybe something that wouldn't be totally messed up by too many complicated safetys, with just altitude holding. Where up means up, down means down, and so on...! I don't know, except for the camera, a bottom-of-the-range Hubsan can do that...

 

[Fracapouille]

At the end of the flight, the aircraft was pointing to the northeast and flying to the right (southeast), and accelerating because you were applying right aileron. It was not ignoring your stick inputs - you were simply not aware of its orientation even though that would have been displayed on your device.

I am affirmative: the drone has gone to crash on a cliff that is situated on the north of its position. It has not gone to the south-east. The gimbal should still be at the bottom of this cliff...
It is described in post #25, with one of the last picture (taken around 800s).

To reiterate, there is no evidence in the logs that the aircraft ever did anything contrary to stick inputs.

Except the location where it crashed. Not regarding its trajectory, but how can you crash on the south face of a cliff, coming from the south, hiting right side first (by visual testimonial), while the drone is pointing northeast?