Mavic 3 flyaway

[aerovikings]

If the drone goes into an automatic RTH because of loss of signal, does pushing the RTH button then cancel RTH?

Yes.

 

[Meta4]

Apparently, AirData only sampled data for the takeoff area; winds could be significantly different closer to the cliffs.

Airdata calculates windspeed by comparing the drine's pitch and roll angles with the drone's speed and directioon.
The wind data posted showed almost no wind at 40 seconds when the drone was at 340 feet above the launch point.
If there was any wind, it would have been detected there and then,

But wind was clearly not a factor in this incident.
A drone will always try to hold position and fight against the wind.
For wind to push a drone at up to 66 mph, the windspeed would have been approximately 100 mph.
It's not possible for a wind that strong to go unnoticed so close to the flyer.
And it's not believable that the wind would trace a neat curved path either.
Wind was not a factor in this incident.
It was a straightforward yaw error, just like a hundred other yaw error incidents that have been posted in the forum previously.

 

[AKAurora]

I may have missed it, but I did not see a clarification on the parking surface, asphalt, concrete, etc.

gravel. All dirt roads and gravel.

 

[AKAurora]

Airdata calculates windspeed by comparing the drine's pitch and roll angles with the drone's speed and directioon.
The wind data posted showed almost no wind at 40 seconds when the drone was at 340 feet above the launch point.
If there was any wind, it would have been detected there and then,

But wind was clearly not a factor in this incident.
A drone will always try to hold position and fight against the wind.
For wind to push a drone at up to 66 mph, the windspeed would have been approximately 100 mph.
It's not possible for a wind that strong to go unnoticed so close to the flyer.
And it's not believable that the wind would trace a neat curved path either.
Wind was not a factor in this incident.
It was a straightforward yaw error, just like a hundred other yaw error incidents that have been posted in the forum previously.

Correct. But DJI is failing to elaborate for me on what exactly the "issue" is. Just that there seemed to be an error and they have sent me a new drone (sitting on my desk, debating selling it because I purchased a replacement figuring I was screwed as I didn't purchase care-refresh.)

 

[Meta4]

Correct. But DJI is failing to elaborate for me on what exactly the "issue" is. Just that there seemed to be an error and they have sent me a new drone (sitting on my desk, debating selling it because I purchased a replacement figuring I was screwed as I didn't purchase care-refresh.)

It looks like you got lucky and DJI analysis made a mistake (that happens sometimes).
The other thing about DJI analysis is that their explanation is usually very poor.

Your incident has been explained several times in the thread, but the message might have been lost between a number of wild guesses that were put forward..
It was caused by a yaw error which happened at power up of the drone

 

[AKAurora]

It looks like you got lucky and DJI analysis made a mistake (that happens sometimes).
The other thing about DJI analysis is that their explanation is usually very poor.

Your incident has been explained several times in the thread, but the message might have been lost between a number of wild guesses that were put forward..
It was caused by a yaw error which happened at power up of the drone, but couldn't cause any trouble until you had GPS reception.

I had gps reception before I got off the ground. I waited until I got 12 satellites and it was no longer orange/yellow and updated the home point.

 

[Meta4]

I had gps reception before I got off the ground. I waited until I got 12 satellites and it was no longer orange/yellow and updated the home point.

Sorry ... I mixed up your incident with a different one.
You had GPS but that you also had the IMU gyro sensor giving inaccurate directional information becaue the compass was affected by magnetic interference at startup.
This caused a yaw error .

After the drone left the startup area, the compass returned to normal, but the IMU can't and the resulting data conflict caused the incident.

 

[AKAurora]

Sorry ... I mixed up your incident with a different one.
You had GPS but that you also had the IMU gyro sensor giving inaccurate directional information becaue the compass was affected by magnetic interference at startup.
This caused a yaw error .

After the drone left the startup area, the compass returned to normal, but the IMU can't and the resulting data conflict caused the incident.

That makes more sense. I've learned a lot and am grateful that I wasn't just out the cost of a new drone.

 

[slup]

And just to give you some more in depth knowledge about yaw errors @AKAurora ...

Just when you power on the drone the IMU (that handles the flight airborne ... not the compass) doesn't know in which direction the drone points. This directional information (heading direction) is just there during the first initial seconds after the power on taken from the compass ... & it "initializes" the IMU yaw heading direction.

So if the compass is magnetically deflected during those initial seconds after power on the IMU yaw will get the same "deflected" heading.

So if we play with this a bit & do a intellectual experiment ...

Let's say that the compass were deflected a half turn (180 degrees) by some ferro magnetic object close enough to the compass during the first seconds after power on ... could be a big steel divers watch on your wrist holding the drone ... or a nail/screw laying in the dirt right below the drone if it's powered on ground ... or rebars hidden in the concrete sidewalk ... or more obvious, the metal table top, or car hood, or metal manhole cover. In many of these examples the magnetic field isn't big enough to trigger a compass calibration prompt ... so you don't notice anything odd at all.

So you launch the drone which in this example in reality is pointing north (... but the IMU think's it points to south).

In this example we have some wind also ... the wind comes from north & is strong enough to push the drone out of position, the drone drifts in reality backwards to the south, so the obvious action to keep position is to speed up the rear motors to pitch nose down to correct the position & hold against the wind.

But as the IMU think's it's pointing nose to the south it notice a drift to the south but here the IMU think's it's drifting forward to south ... the action from the flight controller will here instead be to speed up the front motors making the drone pitch nose up to hold position against the wind. But as the IMU have a 180 degree wrong heading direction towards reality this action will instead increase the positional error instead of bringing the drone back ... so the flight controller tries again, this time even more forceful, but it fails & the positional error just grows ... & it tries again & again, all within milliseconds. The flyaway is a fact & the drone quickly ramps up the speed and in reality flyes rapidly backwards to the south often exceeding both specified tilt and speed limits.

When the yaw error is near 180 degrees the flyaway path is usually straight, but if the error instead is close to 90 degrees the path usually is curved ... just as in your flight.

So what's the take away from this ... can this be avoided?

Firstly ... if the app pop up a compass calibration prompt, always question that. Power down & move to another take off spot far away from the first & power on again, if the prompt comes up again then you can do it.

Secondly ... ALWAYS confirm the drones heading direction on the map view in the app before you goes airborne ... if the drone symbol on the map (the arrow) is pointing equal to other map objects (roads, houses) as the drone does in reality then the IMU yaw initialization have been successful & the IMU have a true heading direction & you're good to go. If the drone direction on the map doesn't match reality ... POWER DOWN, move away, power up & repeat the map check.

Then a bit about DJI's vague information ...

Incidents like yours are embarrasing for them ... they know that this with a deflected compass during power on is a "soft spot" in the system & that they have taken the decision to not explain in a proper way in the users manual what can happen, they just write that you shouldn't calibrate the compass or take off in a magnetically disturbed spot ... nothing more.

So when incidents like yours are filed they usually take their responsibility ... if they can't find something else in order to claim pilot error.

 

[AKAurora]

And just to give you some more in depth knowledge about yaw errors @AKAurora ...

Just when you power on the drone the IMU (that handles the flight airborne ... not the compass) doesn't know in which direction the drone points. This directional information (heading direction) is just there during the first initial seconds after the power on taken from the compass ... & it "initializes" the IMU yaw heading direction.

So if the compass is magnetically deflected during those initial seconds after power on the IMU yaw will get the same "deflected" heading.

So if we play with this a bit & do a intellectual experiment ...

Let's say that the compass were deflected a half turn (180 degrees) by some ferro magnetic object close enough to the compass during the first seconds after power on ... could be a big steel divers watch on your wrist holding the drone ... or a nail/screw laying in the dirt right below the drone if it's powered on ground ... or rebars hidden in the concrete sidewalk ... or more obvious, the metal table top, or car hood, or metal manhole cover. In many of these examples the magnetic field isn't big enough to trigger a compass calibration prompt ... so you don't notice anything odd at all.

So you launch the drone which in this example in reality is pointing north (... but the IMU think's it points to south).

In this example we have some wind also ... the wind comes from north & is strong enough to push the drone out of position, the drone drifts in reality backwards to the south, so the obvious action to keep position is to speed up the rear motors to pitch nose down to correct the position & hold against the wind.

But as the IMU think's it's pointing nose to the south it notice a drift to the south but here the IMU think's it's drifting forward to south ... the action from the flight controller will here instead be to speed up the front motors making the drone pitch nose up to hold position against the wind. But as the IMU have a 180 degree wrong heading direction towards reality this action will instead increase the positional error instead of bringing the drone back ... so the flight controller tries again, this time even more forceful, but it fails & the positional error just grows ... & it tries again & again, all within milliseconds. The flyaway is a fact & the drone quickly ramps up the speed and in reality flyes rapidly backwards to the south often exceeding both specified tilt and speed limits.

When the yaw error is near 180 degrees the flyaway path is usually straight, but if the error instead is close to 90 degrees the path usually is curved ... just as in your flight.

So what's the take away from this ... can this be avoided?

Firstly ... if the app pop up a compass calibration prompt, always question that. Power down & move to another take off spot far away from the first & power on again, if the prompt comes up again then you can do it.

Secondly ... ALWAYS confirm the drones heading direction on the map view in the app before you goes airborne ... if the drone symbol on the map (the arrow) is pointing equal to other map objects (roads, houses) as the drone does in reality then the IMU yaw initialization have been successful & the IMU have a true heading direction & you're good to go. If the drone direction on the map doesn't match reality ... POWER DOWN, move away, power up & repeat the map check.

Then a bit about DJI's vague information ...

Incidents like yours are embarrasing for them ... they know that this with a deflected compass during power on is a "soft spot" in the system & that they have taken the decision to not explain in a proper way in the users manual what can happen, they just write that you shouldn't calibrate the compass or take off in a magnetically disturbed spot ... nothing more.

So when incidents like yours are filed they usually take their responsibility ... if they can't find something else in order to claim pilot error.

Couldn't agree more. It's another thing I've added to my preflight checks because of this incident. I appreciate the explanation and it does make sense.

 

[Apostol.]

BTW @AKAurora the cache file you uploaded can be fixed.
I am sure there are offline/free ways but i just tried all the online repair tools and one of them worked.
Needs payment for the full video but it generated a small preview and it seems to work fine.

https://fix.video/uploads/62e7aebf56c4e4.40650748/fixed.mp4_preview.mp4?394

That would settle a lot of the debates in this thread

 

[AKAurora]

BTW @AKAurora the cache file you upload can be fixed.
I am sure there are offline/free ways but i just tried all the online repair tools and one of them worked.
Needs payment for the full video but it generated a small preview and it seems to work fine.

https://fix.video/uploads/62e7aebf56c4e4.40650748/fixed.mp4_preview.mp4?394

That would settle a lot of the debates in this thread

That's awesome! Thank you very much!

 

[AKAurora]

Still pretty broken, but here it is.

Telluride.mp4

drive.google.com

That's awesome! Thank you very much!