Fly-away and crash at 30mph...

[sar104]

In ATTI mode, the sensors do nothing, above 22 mph, the sensors also do nothing to avoid obstacles, if that is what you are referring to. ATTI has only a barometric sensor for navigation, it will attempt to hold altitude, but that is all. Without GPS data, it would not know how fast is was moving, there is no air speed sensor. In a slip, a plane will accelerate if no opposing input is made, this will result in a high speed slipping movement that can raise the speed significantly in a very short time.

One slight correction - the log file does contain x, y, z velocities even in ATTI with no position data, which the FC is, presumably, calculating by integrating the accelerometer data from the IMU.

 

[Bakersfield Quad]

One slight correction - the log file does contain x, y, z velocities even in ATTI with no position data, which the FC is, presumably, calculating by integrating the accelerometer data from the IMU.

OK, that sounds right, thanks for the clarification. I was just a bit amazed that the OP was not aware of the differences between the modes and how the sensors come into play. Above 22 mph, the sensors do nothing, so at 80mph, of course it would not avoid the cliff.
I think there is a misconception on how the drone works, maybe?

 

[Bakersfield Quad]

I would like to bring up a few points about air, flying and obstacles.
When flying next to cliffs and rock out crops, it is not uncommon to experience lift, turbulence, rotors, waves, chop, up or down drafts.
I have been grabbed by rising air in hang gliders next to cliffs, once was dropped like a stone for over 3,000 feet, with 0 control. Lifted up at over 2,000 feet per minutes until I fought my way out.
I have suffered loss if control in air planes, hang glider and ultralight aircraft, from nothing more than air currents an turbulence.
It would be wise to study micro-meterology if you plan on flying near, cliff, upslopes and such areas. The air is like water, always moving and full of surprises.

 

[sar104]

OK, that sounds right, thanks for the clarification. I was just a bit amazed that the OP was not aware of the differences between the modes and how the sensors come into play. Above 22 mph, the sensors do nothing, so at 80mph, of course it would not avoid the cliff.
I think there is a misconception on how the drone works, maybe?

I think a lot of people are probably unaware of those details. And many are not at all experienced flying in ATTI mode. The Mavic doesn't even have that option.

 

[Fracapouille]

Yes - the yaw/compass heading data are good. It did not rotate and was reacting correctly to rudder input right up until it crashed. It was on a heading of approximately 72° until about 10 seconds before the crash, when you rotated it counterclockwise to a heading of approximately 27°, which it held until it crashed.

View attachment 86643

Thanks you very much for your explanation.
It comforts me with the idea I couldn't make any mistake by pushing the ailerons full right.

Thanks also o Bakersfield for your explanation about the ATTI mode.

A little bit more context below:

In the red window, the situation when the last picture was taken, at around 802s. This picture is below.
Then in grey some short moves of the drone in the area between 802 and 821s.
In yellow, the last actions.
In orange, the attempt to bring the drone back.

I can't see how, at any moment, with a heading at 27° and pushing the aileron full right, I could have crashed the drone like it did. Absolutely no wind at the moment.

Would you confirm the flyaway too?

Thanks again...

 

[Fracapouille]

Sorry I didn't have time enough to check your last messages:

I would like to bring up a few points about air, flying and obstacles.
When flying next to cliffs and rock out crops, it is not uncommon to experience lift, turbulence, rotors, waves, chop, up or down drafts.
I have been grabbed by rising air in hang gliders next to cliffs, once was dropped like a stone for over 3,000 feet, with 0 control. Lifted up at over 2,000 feet per minutes until I fought my way out.
I have suffered loss if control in air planes, hang glider and ultralight aircraft, from nothing more than air currents an turbulence.
It would be wise to study micro-meterology if you plan on flying near, cliff, upslopes and such areas. The air is like water, always moving and full of surprises.

As a fixed-wing drone user since 2012, I am aware of these phenomenons, and I also had a few suprises.
It was a perfectly calm day, not a single breeze anywhere. Really nothing, no movement on any leaves, nada. The drone was only 30m high, just above the trees. It accelerated continuously in direction of the cliff, inclined in this direction. If it was responding, it would at least have incined itself in the opposite direction that I was indicating via the RC, fighting against the wind.

OK, that sounds right, thanks for the clarification. I was just a bit amazed that the OP was not aware of the differences between the modes and how the sensors come into play. Above 22 mph, the sensors do nothing, so at 80mph, of course it would not avoid the cliff.
I think there is a misconception on how the drone works, maybe?

I admit that I'm not an experienced user. I understand that in ATTI the VPS is inefficient, but I didn't know if the lateral sensors where also not activated (and if so, why?).
I didn't know about the 22 mph threshold either. I wouldn't expect it to avoid the cliff of course, but maybe I was expecting it to decrease the speed a little bit before impact. Anyway...

 

[Fracapouille]

I would like to make a precision too: I understand your remarks from a security point of view. The more you know about your drone, the better you can avoid risky situations.
I still have to familiarize with this tool, but one also have to fly to get experienced. I am quite familar with the fixed-wing that is my main tool (400+ flights), and I have just bought a MR a few monthes ago in addition for rare occasions (in my business) where the fixed-wing can't do the job efficiently, and when oblique imagery becomes necessary.

To be honest, I don't enjoy particularly flying those devices. To me, it remains a tool like any other, with their specific constraints of course. I don't fly as a leisure (drone or any other aircraft), I don't fly for other purposes than land surveying, I am not a pilot and I will probably never become one. I am a drone user because these tools open new opportunities for my professsional field. I do my best to avoid any crash, and I would certainly not be there if I didn't have the desire to understand what happened.

That being said, I have chosen a P4P not especially for its technical abilities, but for its reliability, and because it seemed the most handy drone on the market. To me, the easier the better, and this is one of DJI's claims. I have expericenced crashes with my fixed-wing, but I always managed to find out what was faulty (and in general it was me). With this one, I still wonder what happened, and it makes me feel very uncomfortable. Besides the security problem, how much can I rely on this tool, as long as I use it properly? This is what I need to figure out...

 

[sar104]

Thanks you very much for your explanation.
It comforts me with the idea I couldn't make any mistake by pushing the ailerons full right.

Thanks also o Bakersfield for your explanation about the ATTI mode.

A little bit more context below:

In the red window, the situation when the last picture was taken, at around 802s. This picture is below.
Then in grey some short moves of the drone in the area between 802 and 821s.
In yellow, the last actions.
In orange, the attempt to bring the drone back.

I can't see how, at any moment, with a heading at 27° and pushing the aileron full right, I could have crashed the drone like it did. Absolutely no wind at the moment.

View attachment 86670 View attachment 86671

Would you confirm the flyaway too?

Thanks again...

When I said the heading was 27°, I was referring to the direction that the aircraft was pointing. Its track (direction of travel) was reporting very different, 121° just before impact. So it was facing NNE, and traveling ESE (i.e. sideways to the right). Right aileron would be accelerating it.

More worryingly, the log data do not seem to agree well at all with your aerial map. Integrating the inertial velocity data with respect to time to get position relative to takeoff point, we get the following position trace:

That doesn't seem consistent with your map at all, so I'm not sure what to conclude. I cannot see any inconsistencies in the log file data itself, unless I'm missing something that should be obvious. I guess I'll try the same exercise with a known good flight log file to compare inertially derived position with GPS.

 

[Fracapouille]

When I said the heading was 27°, I was referring to the direction that the aircraft was pointing. Its track (direction of travel) was reporting very different, 121° just before impact. So it was facing NNE, and traveling ESE (i.e. sideways to the right). Right aileron would be accelerating it.

More worryingly, the log data do not seem to agree well at all with your aerial map. Integrating the inertial velocity data with respect to time to get position relative to takeoff point, we get the following position trace.

That doesn't seem consistent with your map at all, so I'm not sure what to conclude. I cannot see any inconsistencies in the log file data itself, unless I'm missing something that should be obvious. I guess I'll try the same exercise with a known good flight log file to compare inertially derived position with GPS.

Well, that's weird. At first I thought you might have inverted X and Y coordinates, but it doesn't seem to fit either.
I'm quite confident that the drone crashed going sideways (thus pulling right to compensate), so it seems logical to get something like 90° between track and heading.

My flight path should look like that, the cross is where I took off (and never landed...):

 

[sar104]

Well, that's weird. At first I thought you might have inverted X and Y coordinates, but it doesn't seem to fit either.
I'm quite confident that the drone crashed going sideways (thus pulling right to compensate), so it seems logical to get something like 90° between track and heading.

My flight path should look like that, the cross is where I took off (and never landed...):View attachment 86679

Yes - I spent some time look at the coordinate scheme, but that appears to be correct. Did you have the drone in sight for the entire flight, or enough of the flight to be confident where it was?

In terms of the sideways crash that seems consistent, but presumably you thought it was flying to the left if you applied right aileron to try to stop it. The recorded motion and pitch data seem to indicate clearly that it was flying to the right.

 

[Fracapouille]

Yes - I spent some time look at the coordinate scheme, but that appears to be correct. Did you have the drone in sight for the entire flight, or enough of the flight to be confident where it was?

Yes, I had to keep the drone in sight all along the flight, it is mandatory in those conditions. Plus I have all the pictures of said flight, as I was doing a photogrammetric survey. I can send them to you if you need them.

The recorded motion and pitch data seem to indicate clearly that it was flying to the right.

This.

Does it rely on the compass in the ATTI mode to know how it is orientated? Could a magnetic inversion be the cause?

 

[Fracapouille]

Could it be a lead?

 

[BudWalker]

@sar104 , @Fracapouille if you could retrieve the .DAT we could know more about possible compass issues. Also, the IMU data will indicate which side of the AC impacted the cliff. It's also possible there may be some GPS data, flawed though it may be. If the P4 can be turned on then you could use these instructions
How to retrieve a .DAT
to retrieve the .DAT. If it can't be turned on then you would have to extract the internal SD card to get the .DAT

 

[sar104]

Yes, I had to keep the drone in sight all along the flight, it is mandatory in those conditions. Plus I have all the pictures of said flight, as I was doing a photogrammetric survey. I can send them to you if you need them.



This.

Does it rely on the compass in the ATTI mode to know how it is orientated? Could a magnetic inversion be the cause?

The finer details of the FC programming are unknown, I think, at least in terms of which sensors are used and prioritized in which situations. If you look at the log file for an ATTI flight such as this one, it includes X, Y and Z velocity data (horizontal frame of reference), pitch and roll angles, and yaw angle. The yaw angle is the compass bearing that the aircraft is facing. Pitch and roll angles are obviously relative to the aircraft. By inspection of the log files from GPS flights it appears that the X, Y velocity data are relative to north (presumably as determined by compass data), with positive X as north. In ATTI mode velocity V(X, Y) can only be calculated by integrating the x, y, z accelerometer data (aircraft frame of reference) to give v(x, y), and then transforming that to the X, Y frame of reference by projecting to the horizontal plane and then calculating the N and E velocity components in that plane.

In theory, all the FC needs to know is the compass bearing at takeoff and then integration of the inertial sensors will yield linear and rotational motion. Or the rotational motion could be derived purely from compass data, or a combination of the two.

I guess it is also possible that the X, Y frame of reference is different in ATTI but, with compass data to lock north, I can't see why that would be the case.

I think I need to put my P4 through some controlled GPS and ATTI flight paths to see if I can confirm those assumptions. Sounds like a job for the weekend.

 

[Fracapouille]

@sar104 , @Fracapouille if you could retrieve the .DAT we could know more about possible compass issues. Also, the IMU data will indicate which side of the AC impacted the cliff. It's also possible there may be some GPS data, flawed though it may be. If the P4 can be turned on then you could use these instructions
How to retrieve a .DAT
to retrieve the .DAT. If it can't be turned on then you would have to extract the internal SD card to get the .DAT

Here you are!
Dropbox - FLY034.DAT

Thanks a lot!

 

[chris digata]

if i have to take pff in atti with no gps, then i certainally keep in in atti for the rest of the flight

 

[BudWalker]

Here you are!
Dropbox - FLY034.DAT

Thanks a lot!

At the very end of the flight there was some compass anomaly. But, it was very small and didn't have any significant effect. The magnetometers were making it look like there was CCW rotation. But, totalGyroZ (computed by integrating and summing gyroZ) didn't show any rotation. This might have been a problem if the flight hadn't been stopped short by the impact.

The velocity and Yaw data show the heading was roughly 36 degrees and the direction of travel 128 degrees. Also, full right rudder was being applied. I believe this is consistent with @sar104 's analysis. BTW, in this flight the Yaw and velocity values will be magnetic, not true. This is because declination is computed from the GPS data which wasn't available.

The accelerometer data is confusing and doesn't present a clear picture of the impact. At least I find it confusing. I always have to remember what positive and negative accelerations mean in terms of an impact. Initially, there is relatively small negative X and Y acceleration indicating the impact was in the left rear quadrant. This is followed by a much higher positive X acceleration indicating impact in the front. I've attached a .zip containing a .csv for the last 4 secs and at the highest sampling rate possible. Maybe someone can make some sense of this.

Unfortunately, there was no GPS data.

 

[sar104]

At the very end of the flight there was some compass anomaly. But, it was very small and didn't have any significant effect. The magnetometers were making it look like there was CCW rotation. But, totalGyroZ (computed by integrating and summing gyroZ) didn't show any rotation. This might have been a problem if the flight hadn't been stopped short by the impact.

The velocity and Yaw data show the heading was roughly 36 degrees and the direction of travel 128 degrees. Also, full right rudder was being applied. I believe this is consistent with @sar104 's analysis. BTW, in this flight the Yaw and velocity values will be magnetic, not true. This is because declination is computed from the GPS data which wasn't available.

The accelerometer data is confusing and doesn't present a clear picture of the impact. At least I find it confusing. I always have to remember what positive and negative accelerations mean in terms of an impact. Initially, there is relatively small negative X and Y acceleration indicating the impact was in the left rear quadrant. This is followed by a much higher positive X acceleration indicating impact in the front. I've attached a .zip containing a .csv for the last 4 secs and at the highest sampling rate possible. Maybe someone can make some sense of this.
View attachment 86697

Unfortunately, there was no GPS data.

Interesting discrepancy between the DAT file yaw and the log file yaw data in the seconds leading up to impact. Is that variation normal?

 

[sar104]

Here you are!
Dropbox - FLY034.DAT

Thanks a lot!

So, just to clarify, at the end of the flight, before it crashed, can you confirm that the drone was heading north, or a little west of north, as indicated in your diagrams? That is certainly consistent with the location you marked for the crash, but it is in completely the opposite direction from the log and DAT data record.

In terms of the accuracy of the inertial-data-derived track that I put in post #28, that is in agreement with your estimated position at 800 s, but apparently diverges wildly in the final 30 s from where it ended up.

Also - if you can give coordinates for the location then I can superimposed the derived track on the aerial image for comparison.

 

[Fracapouille]

The velocity and Yaw data show the heading was roughly 36 degrees and the direction of travel 128 degrees. Also, full right rudder was being applied. I believe this is consistent with @sar104 's analysis. BTW, in this flight the Yaw and velocity values will be magnetic, not true. This is because declination is computed from the GPS data which wasn't available.

The accelerometer data is confusing and doesn't present a clear picture of the impact. At least I find it confusing. I always have to remember what positive and negative accelerations mean in terms of an impact. Initially, there is relatively small negative X and Y acceleration indicating the impact was in the left rear quadrant. This is followed by a much higher positive X acceleration indicating impact in the front. I've attached a .zip containing a .csv for the last 4 secs and at the highest sampling rate possible. Maybe someone can make some sense of this.

Thanks for your participation in the discussion. Heading of 36°, and direction of travel of 128° are not compatible with an impact on the left side of the drone, and with how things happened. Those values are can't be correct, and there is definitely something wrong here...
But I still don't get if and how this anomaly may have disrupted the attitude of the drone... I mean, trivially, any problem there might be regarding the referential, the left, right, front, back sides of the drone remain the same, and when using the controls, I don't see why the referential would interfere there? Except for yaw...
If this asumption is true, do you agree this means:
- either this anomaly may, or may not have lead to the crash, but anyway I didn't have control at all on the drone anymore and it decided itself what to do. Meh...
- or the drone was effectively accelerating and hit rightwards, and then there must have been a 180° turn at some time. If so, there should be a record of that command between the moment when the last picture was taken and the time it crashed. If there's no such record, it means at some time I lost control of the drone too...
In that case there is a good chance the compass anomaly was the reason it lost its orientation. I can't confirm that this U-turn has occurred, it can be really fast. There is something that could confirm this scenario: I took pictures of the drone before sending it to repair. It was badly damaged (TBH, I didn't want to look at this mess too much, it was quite depressing), but it seems the right arms have been more damaged:

Are you affirmative about the left side impacting first?

Thanks to all of you for your implication...