Mavic Pro 2 Motor Failure : crash and lost. What to do ?

[Prismatic]

Your last sentence. I'm reading all this and thinking the same. How big the birds.
I'm intrigued by how knowledgeable members here in the forum are. ??

At this point it seems virtually certain that the first event, at least, was a bird strike. Subsequent events appear to be less obvious, and even the experts are less than 100% sure of exactly why the drone went out of control.

Anyway, I think birds usually attack for one of three reasons: nest defense, territorial defense, or predation. Aerial predators tend to be larger species, and I'd guess the drone was hit by a hawk or eagle, or something similar. The intent of the bird--defense or offense--can't be known with certainty, but the attack angle--from above--suggests a predator aiming for a kill. Regardless, it was a big, heavy bird: no finch or swallow is going to flip a Mavic 2!

I applied a dazzling, fiery-looking holographic skin to my Mavic Air, hoping that such an "unnatural" appearance will deter birds from attacking (in either defense or offense!). So far, so good, but one bird-strike would prove me wrong.
?

Standing before the Golden Drone -- a bird averting skin

Birds fear and flee from fire. The dazzling golden holographic skin of my drone, Bertie, flashes hues of red and orange, as if on fire. It's intended to alarm an approaching bird, and make it break away to safety. The photo and video only suggest the extreme visual impact of Golden Bertie in...

mavicpilots.com

 

[sar104]

Doesn't the acceleration data show that the M2 was struck from the left, back and above? I believe that the negative accelX means the AC accelerated forward along the X axis.
View attachment 90949
Also, the M2 was stationary when it was hit - almost certainly a bird strike.

But, the impact point was the right front quadrant as can be seen from the pitch and roll data.
View attachment 90950
When the M2 was hit from above in the right front quadrant it rolled CW and pitched down.

No - I don't think so. In the earth frame of reference, north is x, east is y and down is z. In the aircraft frame of reference, forward is x, right is y, and down is z. Anything else would be an inconsistent use of the xyz notation, as well as out of convention. So positive x-axis acceleration is forward. That then defines yaw as CW from north, as expected.

​

At the first impact, the aircraft was stationary and had made several small CW rotations, leaving it facing WNW (-67°). The impact drove it backwards (SE):



Looking at the accelerations (aircraft FOR), recorded north and east velocities, together with those velocities transformed into the aircraft frame of reference, and the yaw:



That's consistent the FOR definitions above - the accelerations are negative in x and y, and the aircraft is kicked backwards and left. That puts the impact front right, also consistent with the motor data.

 

[sar104]

Think I agree with @BudWalker regarding accelX direction ... forward movement creates minus values.

I think you are neglecting pitch excursions that generally accompany forward/backward motion. To accelerate forwards the aircraft pitches forwards, which changes the orientation of the x-axis accelerometer to point downwards. In level flight there is a constant negative 1 g downward acceleration, and so the accelerometer sees a component of that, which is actually larger than the component of forward acceleration that it detects. Note that in regular flight the negative x-axis acceleration recorded continues even when the aircraft reaches constant forward velocity, confirming that it is due to pitch, not horizontal acceleration.

 

[slup]

I think you are neglecting pitch excursions that generally accompany forward/backward motion. To accelerate forwards the aircraft pitches forwards, which changes the orientation of the x-axis accelerometer to point downwards. In level flight there is a constant negative 1 g downward acceleration, and so the accelerometer sees a component of that, which is actually larger than the component of forward acceleration that it detects. Note that in regular flight the negative x-axis acceleration recorded continues even when the aircraft reaches constant forward velocity, confirming that it is due to pitch, not horizontal acceleration.

@sar104, a bit free from the thread event here below ...
I'm following your reasoning, so are you saying that just looking at accelX & accelZ separately we get (I get ...) tricked, this as with an ongoing pitch which turns the coordinate system forward it's actually a accelComposite XZ that tells the real story?

Or do I miss the point ..?

This is how it looks for a Mavic Air in Sport with full positive elevator from stationary ... the pitch forward down to max 35 degrees is nearly instant, but the accelX continues much longer & reaches it's most negative value about the same time as the velocity start to reach it's maximum.

 

[Tristan29photography]

I feel sadly for your loss. I hope things will look up soon with a new drone for you. ?

Thank you so much. Like we say in Colombia, "Adelante Pues". Ahead, tomorrow is always better

 

[Tristan29photography]

Hi Guys,

So I have been this morning looking for the M2 with 2 local guides.
We have been searching for 2h again in the jungle, and didn't find anything. It could be stuck at the top of the trees, or have fallen into the river... We've even been to the closest indigenous village to ask if they had found a drone, but they didn't know what was a drone ;-)

So that's finished.

In memory, here is the last video cached before crashing.

Thank you all for your help and diagnostics !

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

 

[sar104]

@sar104, a bit free from the thread event here below ...
I'm following your reasoning, so are you saying that just looking at accelX & accelZ separately we get (I get ...) tricked, this as with an ongoing pitch which turns the coordinate system forward it's actually a accelComposite XZ that tells the real story?

Or do I miss the point ..?

This is how it looks for a Mavic Air in Sport with full positive elevator from stationary ... the pitch forward down to max 35 degrees is nearly instant, but the accelX continues much longer & reaches it's most negative value about the same time as the velocity start to reach it's maximum.

View attachment 90999

Yes. What you need to do to visualize it is to convert the accelerations in the aircraft frame of reference into accelerations in the earth's frame of reference - most easily done by working directly through the orientation quaternions. I'll do an example this evening if I get time. Or if you want to post the data from your graph above I can use that directly for comparison.

 

[BudWalker]

No - I don't think so. In the earth frame of reference, north is x, east is y and down is z. In the aircraft frame of reference, forward is x, right is y, and down is z. Anything else would be an inconsistent use of the xyz notation, as well as out of convention. So positive x-axis acceleration is forward. That then defines yaw as CW from north, as expected.

View attachment 90992​

At the first impact, the aircraft was stationary and had made several small CW rotations, leaving it facing WNW (-67°). The impact drove it backwards (SE):

View attachment 90994

Looking at the accelerations (aircraft FOR), recorded north and east velocities, together with those velocities transformed into the aircraft frame of reference, and the yaw:

View attachment 90995

That's consistent the FOR definitions above - the accelerations are negative in x and y, and the aircraft is kicked backwards and left. That puts the impact front right, also consistent with the motor data.

Ok, didn't quite follow the first part. But, I did do a test with my Mavic Pro. It was held steady and then jerked like it had been hit from above, right and forward. That matches the accelerations seen here.

As you said negative acceleration means that the AC was accelerating in the negative direction.

Like @slup I had been fooled by the X axis dipping down into the constant -1 g from the earth's gravitational field. Some may find this plot interesting.

In the left half the M2 was hovering and being slowly rotated. The oscillating pitch is being caused by the M2 tilting into the wind to remain in place. Note that the accelX is oscillating in phase with the pitch as the accelX dips in and out of the -1 g gravitational filed.

In the 2nd half full elevator has been applied and the pitch is about -9.6° causing the accelX to dip into the -1 g gravitational filed. tan(-9.6° ) is -0.169 which matches the measured accelX.

 

[sar104]

Ok, didn't quite follow the first part. But, I did do a test with my Mavic Pro. It was held steady and then jerked like it had been hit from above, right and forward. That matches the accelerations seen here.
View attachment 91005
As you said negative acceleration means that the AC was accelerating in the negative direction.

Like @slup I had been fooled by the X axis dipping down into the constant -1 g from the earth's gravitational field. Some may find this plot interesting.
View attachment 91008
In the left half the M2 was hovering and being slowly rotated. The oscillating pitch is being caused by the M2 tilting into the wind to remain in place. Note that the accelX is oscillating in phase with the pitch as the accelX dips in and out of the -1 g gravitational filed.

In the 2nd half full elevator has been applied and the pitch is about -9.6° causing the accelX to dip into the -1 g gravitational filed. tan(-9.6° ) is -0.169 which matches the measured accelX.

The first bit was just an observation of the mathematical implications of the relative axis orientations in the cartesian coordinate system that follow the right hand rule.

 

[slup]

...Or if you want to post the data from your graph above I can use that directly for comparison.

Much appreciated @sar104 Will be interesting reading tomorrow, or when you have the time (Sweden is like 8 hour before you, so now I need to hit the bed...).

My DAT attached ... You see the time window in my chart earlier.

?

 

[sar104]

Much appreciated @sar104 Will be interesting reading tomorrow, or when you have the time (Sweden is like 8 hour before you, so now I need to hit the bed...).

My DAT attached ... You see the time window in my chart earlier.

?

Here are the aircraft FOR accelerations transformed into horizontal acceleration:



As you noted, the x-axis acceleration (red) goes negative and stays there, even though the aircraft velocity becomes constant. But that's caused ay the negative pitch (green). The actual horizontal forward acceleration (blue) goes positive as the aircraft accelerates forwards, and then returns to zero. Then at 380 seconds, the pitch goes positive to stop the aircraft, the x-axis acceleration goes back towards zero, but the actual horizontal acceleration goes negative, as expected.

 

[slup]

Here are the aircraft FOR accelerations transformed into horizontal acceleration:

View attachment 91027

As you noted, the x-axis acceleration (red) goes negative and stays there, even though the aircraft velocity becomes constant. But that's caused ay the negative pitch (green). The actual horizontal forward acceleration (blue) goes positive as the aircraft accelerates forwards, and then returns to zero. Then at 380 seconds, the pitch goes positive to stop the aircraft, the x-axis acceleration goes back towards zero, but the actual horizontal acceleration goes negative, as expected.

Thank's @sar104 for taking the time ... will chew a bit below on this & see if I full grasp it.

I understand the difference between the horizontal and X-axis accel & also that (what I think at least) the tilting forward (increasing negative pitch angle) will add negative values to accelX from a negative directed X vector from the down moving (positive moving) Z-axis. This is, as I understand what makes the accelX graph move to the negative side.

This is mainly what tricked me thinking that accel forward is negative, stands corrected & a bit wiser ... looking at both the red & blue graphs, am I correct by assuming that the correct read off time slot in your chart regarding accel differencies should be from just around 369sec to 380sec ?

Then lastly ... is it a way to generate the horizontal accel graph in CsvView?

 

[sar104]

Thank's @sar104 for taking the time ... will chew a bit below on this & see if I full grasp it.

I understand the difference between the horizontal and X-axis accel & also that (what I think at least) the tilting forward (increasing negative pitch angle) will add negative values to accelX from a negative directed X vector from the down moving (positive moving) Z-axis. This is, as I understand what makes the accelX graph move to the negative side.

This is mainly what tricked me thinking that accel forward is negative, stands corrected & a bit wiser ... looking at both the red & blue graphs, am I correct by assuming that the correct read off time slot in your chart regarding accel differencies should be from just around 369sec to 380sec ?

Could you clarify what you are asking? The horizontal acceleration will differ from x-axis acceleration whenever pitch is not zero.

Then lastly ... is it a way to generate the horizontal accel graph in CsvView?

I don't think so, although @BudWalker could implement it. Horizontal forward acceleration, aF, is given by:

aF = aX*cos(pitch) + aY*(sin(roll)*sin(pitch) - cos(roll)) + aZ*(sin(roll) + cos(roll)*sin(pitch))​

 

[slup]

Could you clarify what you are asking? The horizontal acceleration will differ from x-axis acceleration whenever pitch is not zero.


I don't think so, although @BudWalker could implement it. Horizontal forward acceleration, aF, is given by:

aF = aX*cos(pitch) + aY*(sin(roll)*sin(pitch) - cos(roll)) + aZ*(sin(roll) + cos(roll)*sin(pitch))​

Well yeah ... perhaps I got caught in my own fuzzy thinking just before bedtime in the beginning of my previous post

What I'm trying to grasp is what factors from the pitching coordinate system that pushes the accelX to the negative side & makes it stay there even if there isn't any velocity involved?

Below chart is from a test I did with a stationary Mavic Air ... made the pitch negative by hand & then the accelX goes negative and the accelZ goes positive ... both stay there without any movement until I level the AC again.

 

[new2mavic]

What I'm trying to grasp is what factors from the pitching coordinate system that pushes the accelX to the negative side & makes it stay there even if there isn't any velocity involved?

Imagine that you are sitting in a stationary car.
If the car is on a flat road, you can only feel pressure on you butt, ie, accelZ = 1(G), accelX = accelY = 0
If the car is on an inclined road with the front pointing up hill, you will feel the pressure on you butt and your back. The accel values will be something like accelZ = 0.9, accelX = 0.2, accelY = 0.
That's why the log of your MA showed a non-zero accelX when tilted and not moving.

Now imagine that the car is accelerating forward on a level road. You will feel the usual amount of pressure on your butt as well as the seat pushing against your back. So the accel values will be like accelZ = 1G, accelX = 0.1G, accelY = 0.

Hope it helps to explain what accelerometers measure.

 

[slup]

Imagine that you are sitting in a stationary car.
If the car is on a flat road, you can only feel pressure on you butt, ie, accelZ = 1(G), accelX = accelY = 0
If the car is on an inclined road with the front pointing up hill, you will feel the pressure on you butt and your back. The accel values will be something like accelZ = 0.9, accelX = 0.2, accelY = 0.
That's why the log of your MA showed a non-zero accelX when tilted and not moving.

Now imagine that the car is accelerating forward on a level road. You will feel the usual amount of pressure on your butt as well as the seat pushing against your back. So the accel values will be like accelZ = 1G, accelX = 0.1G, accelY = 0.

Hope it helps to explain what accelerometers measure.

Thank's, that picture made it into my mind. Pretty much what I in a somewhat confused state came to yesterday ...with speed forward, a tilting Z-axis start to effect the X-axis in a negative way (falling into the accel direction) & by that move the whole accelX into negative side + at the same time due to a tilting X-axis starts to effect the Z-axis in a positive direction (butt lifts up from the car seat) making the accelZ park itself more positive.

Yeah ... I know, I'm fond of complicating thing's

 

[rewizard]

Perhaps if you could find someone else with a drone you fly similar pattern and find it from above

 

[AntDX316]

Is there anyway to see what happened in 3D over time with the correct altitude? All that data exist so we can actually tell tilt angles orientation and other things? Interpreting with mental math may generate mistakes unless of course the sensors are wrong.

 

[lrsn]

I'm really sorry about what happened to you as it's always my biggest fear! Years ago I lost my new at the time P4 due to a tree top encounter! Just the day before I had received a Marco Polo Locater that sends out it's own signal without internet and without reading the instructions I tie wrapped the Tag to one skid and flew it over a ridge where it started to climb but flew into the top of a 150' tree and fell to the ground in extremely hard to get to terrain up at 8,000' ASL in the Rocky Mountains. After realizing that my just purchased new P4 was not going to return I turned on my Marco Polo Locater and finally received a signal! This was in extremely rough mountines terrain but after walking for 45 minutes I found it! Every one of my drones is now set up with a Marco Polo. Good luck!

 

[AMann]

Your last sentence. I'm reading all this and thinking the same. How big the birds.
I'm intrigued by how knowledgeable members here in the forum are. ??

Columbia has sloth hunting harpy eagles that can even kill people. They grow big birds down there!