Suddenly flew away in high speed and crashed into water

Labroides · 2021-10-25

Johnny_J Posted at 10-25 07:27
There should be an option at least, to select "Do not fly away even if you (the drone) believe you are disoriented"
It's incredibly stupid of DJI to assume that their "intelligence" is superior to the pilot in-place!

That's not possible.
Read my subsequent posts for explanation.

Sean-bumble-bee · 2021-10-25

Thanks for the first part.

The problem isn't in the drone's ability to know it's position.
The problem is having a conflict between the directional data from the compass and gyro sensors.
The drone knows its position, but when it tries to move to correct itself, the trouble begins because the gyro sensors have been initialised with incorrect directional information.

I think you misunderstood my post, I was referring to, and comparing, flights made by me where everything was functional, with no conflicts, but in which there was sufficient light for the VPS to work in one flight (a day time flight) and insufficient light for the VPS to work in the other flight (a night time flight).

Both flights had adequate GPS counts and working compasses etc. but, comparatively speaking, the drone drifted/wandered 'much' more during the night time flight than it did during the day time flight.

I take this as evidence that, when available, the VPS positioning over rides GPS positioning.

mkh · 2021-10-25

Labroides Posted at 10-25 13:40
As far as I remember the compass reading in the app display was not showing any anomaly.
Because there was no anomaly that the compass could detect.
The anomaly was in the direction that the magnetic interference rather than the strength of the magnetic field.

I cant think of any magnetic interference there. It was a flat land near the water body without any metal object / power line or anything which can cause magnetic interference.

Labroides · 2021-10-25

mkh Posted at 10-25 15:52
I cant think of any magnetic interference there. It was a flat land near the water body without any metal object / power line or anything which can cause magnetic interference.

If it was an external source, it would have been within just a few feet or even inches.

Something definitely was affecting the compass before you launched and it got further away from the cause of the problem.

Did you power on where you launched?

DJI Stephen · 2021-10-25

mkh Posted at 10-24 18:02
I have created a case .. https://repair.dji.com/mycase/casedetails?CaseNo=CAS-7492092-G9P1T1

Thanks.

Hi there mkh. Thank you for the reply and it is great to know that you have already contacted out DJI support team and have a case number for your DJI Mini 2. If you have any other concerns or any inquiries with DJI. Please feel free to post it here at DJI Forum. We are all here to help and assist you. Thank you.

JJB* · 2021-10-25

Sean-bumble-bee Posted at 10-25 14:57
Thanks for the first part.

Hi Sean-bumble-bee,

In the log is seen when the VPS positioning is On or Off.
Always Off when GPS is receiving data, turnes ON when GPS is not available.
ON means that only VPS pos is used for drone positioning.
But with GPS the VPS pos is an aid to control its position.

cheers
JJB

JJB* · 2021-10-25

Pleomax Posted at 10-25 11:54
The zip contains a simulation of your flight from the log data. Notice the difference between Gimbal Yaw (360) and the Mini2 Yaw(360).

Hi Pleomax,

Nicely made i am a fan of VB in Excel or did you made it in other program?

cheers
JJB

JJB* · 2021-10-26

Labroides Posted at 10-25 13:40
As far as I remember the compass reading in the app display was not showing any anomaly.
Because there was no anomaly that the compass could detect.
The anomaly was in the direction that the magnetic interference rather than the strength of the magnetic field.

Hi Labroides,

Your saying that a "gyro sensor has been initialised with incorrect data" is confusing for me, as a gyro is an output giver and not a input storer.

In the IMU you have accelerometers and gyro for each axis.
Accelerometer : sensing the acceleration in a axis, so only during speed increase/decrease, linear speed moving is not sensed
Gyro : gives zero value as signal when not turned in that axis, but gives signal during a rotation in that axis.

During IMU initialisation the compass heading is stored in 2 channels, yawing a drone in flight will change both channel values.
1) simply the compass heading change, 2) calculated changed heading, using accelerometer and gyro data.
If both values are the same after the change in heading ; no problem.

If the compass is interfered during initialisation both values in the 2 channels are equal, as the primary heading data is derived from one source ; the compass.
After takeoff, compass away from interference, the compass heading is changing while drone is normally steady in hover.
Or amount of compass heading change does not equal the calculated heading change using acc meter and gyro.

Because the acc meter and gyro does not sense a movement or movenent is less than compass hdg change (while compass hdg is changing away from interference) this value in the other channel is not changed.
So there is a difference in values between channels, yaw error resulting in a fly away.

If you think is it totally wrong, please do explain why (my adage 'learning all the time')

cheers
JJB

droneriffledchi · 2021-10-26

Labroides Posted at 10-25 19:10
If it was an external source, it would have been within just a few feet or even inches.

Something definitely was affecting the compass before you launched and it got further away from the cause of the problem.

Would a watch on the hand which was used to launch the drone could have caused similar problem as described in this thread?

Johnny_J · 2021-10-26

Labroides Posted at 10-25 13:50
That's not possible.
Read my subsequent posts for explanation.

Of course it's possible!
Which subsequent posts? Compass disturbance?

Pleomax · 2021-10-26

JJB* Posted at 10-25 22:57
Hi Pleomax,

Nicely made i am a fan of VB in Excel or did you made it in other program?

Thanks JJB,

Yes all done with VB in Excel. I like to see data update in real time as flight progresses.

Would like to distribute it as an .exe but don't think it's possible with Excel.

What were your thoughts on different values between Gimbal Yaw and Mini2 Yaw?

Looks like Gimbal Yaw was accurate.

Labroides · 2021-10-26

Johnny_J Posted at 10-26 04:59
Of course it's possible!
Which subsequent posts? Compass disturbance?

What's not possible is what you suggested:
There should be an option at least, to select "Do not fly away even if you (the drone) believe you are disoriented"
It's incredibly stupid of DJI to assume that their "intelligence" is superior to the pilot in-place!

I went to some trouble to explain why and I'm not going to type it all out again.
If it was so simple, DJI would have sorted it out years ago, .. but it isn't.
As you'd understand if you read what I wrote.

It's amazing that people with little technica lknowledge are so confident that they know better than DJI's design engineers.

Labroides · 2021-10-26

JJB* Posted at 10-26 00:07
Hi Labroides,

Your saying that a "gyro sensor has been initialised with incorrect data" is confusing for me, as a gyro is an output giver and not a input storer.

The gyro gets its initial directional reference from the compass.

I can't follow what you wrote.

Labroides · 2021-10-26

droneriffledchi Posted at 10-26 01:19
Would a watch on the hand which was used to launch the drone could have caused similar problem as described in this thread?

I doubt it, but do some testing with a magnetic compass and see if teh watch can deflect the compass at a similar distance.

Johnny_J · 2021-10-26

Labroides Posted at 10-26 05:13
What's not possible is what you suggested:
There should be an option at least, to select "Do not fly away even if you (the drone) believe you are disoriented"
It's incredibly stupid of DJI to assume that their "intelligence" is superior to the pilot in-place!

Labroides wrote: "I went to some trouble to explain why and I'm not going to type it all out again"

I asked you what posts in this thread you were referring to.

Labroides wrote: "It's amazing that people with little technica lknowledge are so confident that they know better than DJI's design engineers"

Don't group me with "people with little technical knowledge".
Most people here seem to believe that as soon as there is a metal nearby the drone, the compass will go bananas!
On a boat where they also use (magnetic) compasses, the proximity to metal is unimportant . What is important is the amount of magnetism that metal holds.
The same goes for our drones. Most of the people here will not believe this fact. Hence the downvote!

mkh · 2021-10-26

Johnny_J Posted at 10-26 05:37
Labroides wrote: "I went to some trouble to explain why and I'm not going to type it all out again"

I asked you what posts in this thread you were referring to.

Some time common sense are not so common. Enabling a check before Auto Take off should be there. Dropping the drone from 1.2 Meter to the ground is much much better (also safer) option than high speed fly away to correct the attitude at low altitude / near ground.

Johnny_J · 2021-10-26

mkh Posted at 10-26 06:00
Some time common sense are not so common. Enabling a check before Auto Take off should be there. Dropping the drone from 1.2 Meter to the ground is much much better (also safer) option than high speed fly away to correct the attitude at low altitude / near ground.

Common sense would be to not using "auto take off". Minimize the actions the software can do by itself!

JJB* · 2021-10-26

Labroides Posted at 10-26 05:14
The gyro gets its initial directional reference from the compass.

I can't follow what you wrote.

The only input to a gyro is its power supply, gyro gives an output signal it moves in its direction.
Sorry that you cannot follow my writing, this is how it works.

cheers
JJB

mkh · 2021-10-26

Johnny_J Posted at 10-26 06:11
Common sense would be to not using "auto take off". Minimize the actions the software can do by itself!

Are you suggesting that DJI recommends not to use the Auto Take off and rely on the manual take off? Are you representing DJI here?
Moreover, the all the controls were unresponsive for some reason here. So there was not any option for manual control.

JJB* · 2021-10-26

mkh Posted at 10-26 06:14
Are you suggesting that DJI recommends not to use the Auto Take off and rely on the manual take off? Are you representing DJI here?

No difference in how to take off, if the yaw/compass signal is incorrect a drone will fly-away once it needs correction to hold a stabilized position.

mkh · 2021-10-26

Any idea how long DJI will take to resolve this case?

mkh · 2021-10-26

Are this type of incidents of sudden flyaway without pilot input is very common? Is it related with Mini 2 only? How about Air 2 s?

Johnny_J · 2021-10-26

mkh Posted at 10-26 06:14
Are you suggesting that DJI recommends not to use the Auto Take off and rely on the manual take off? Are you representing DJI here?
Moreover, the all the controls were unresponsive for some reason here. So there was not any option for manual control.

I'm suggesting you read my post again! Minimize DJI Software Involvement!

mkh · 2021-10-26

Johnny_J Posted at 10-26 06:36
I'm suggesting you read my post again! Minimize DJI Software Involvement!

With my limited knowledge and understanding, pretty much all the movement of a quadcopter is controlled by varying the power supplied to the motors, which is determined by the software algo using different sensors inputs. So in any case, we are relying on the software only.

JJB* · 2021-10-26

mkh Posted at 10-26 06:21
Are this type of incidents of sudden flyaway without pilot input is very common? Is it related with Mini 2 only? How about Air 2 s?

It can happens to all DJI drones....

That is why it is so important to check the area where your takeoff your drone

AND

always check before take-off the blue arrow in the fly app map view and compare heading to the actual heading. very basic rule : if not the same > do not fly.

cheers
JJB

Johnny_J · 2021-10-26

JJB* Posted at 10-26 06:14
The only input to a gyro is its power supply, gyro gives an output signal it moves in its direction.
Sorry that you cannot follow my writing, this is how it works.

The main input to a gyro is the change of orientation. You don't need a power supply for that.
A power supply is optional but does not change the physical laws.

JJB* · 2021-10-26

Pleomax Posted at 10-26 05:13
Thanks JJB,

Yes all done with VB in Excel. I like to see data update in real time as flight progresses.

Hi

Same here, so i use FRAP ;-)

cheers
JJB

JJB* · 2021-10-26

Johnny_J Posted at 10-26 07:20
The main input to a gyro is the change of orientation. You don't need a power supply for that.
A power supply is optional but does not change the physical laws.

Hi Johnny,

Great, you found a way to get free energy, a gyro spinning fast on its own. Optional power?
Nah check detailed wire diagrams for electronics.

Main input is not the change of orientation, well depends how you define input....
Input > gyro > Output plus a arrow to the gyro representating the sensing of orientation. no sensing = no output, while input is always there to get that thing working.

cheers
JJB

Johnny_J · 2021-10-26

JJB* Posted at 10-26 07:55
Hi Johnny,

Great, you found a way to get free energy, a gyro spinning fast on its own. Optional power?

Wrong again! A gyro is trying to keep it's position by means of a fast spinning disc with substantial weight.
Whether it is spun up manually by human force or by a motor does not change the behavior of the gyro.
Without saying it, you assume that the gyro will spin forever. That is not true. Either the disc eventually stop due to friction or it keeps on spinning powered by whatever.

mkh · 2021-10-26

JJB* Posted at 10-26 07:13
It can happens to all DJI drones....

That is why it is so important to check the area where your takeoff your drone

As far as I remember the compass was pointing in the same direction. Also it usually complains and instructs to re-calibrate when it detects the anomaly during the start up.

In this case, there was no indication of the compass error.

Sean-bumble-bee · 2021-10-26

JJB* Posted at 10-25 22:53
Hi Sean-bumble-bee,

In the log is seen when the VPS positioning is On or Off.

That's interesting, I will have a look in my logs. Thanks

Sean-bumble-bee · 2021-10-26

decided what was here contributed nothing

JJB* · 2021-10-26

Johnny_J Posted at 10-26 08:09
Wrong again! A gyro is trying to keep it's position by means of a fast spinning disc with substantial weight.
Whether it is spun up manually by human force or by a motor does not change the behavior of the gyro.
Without saying it, you assume that the gyro will spin forever. That is not true. Either the disc eventually stop due to friction or it keeps on spinning powered by whatever.

well i hope that my DJI gyros will spin forever, well for sure i hope for the entire flight.....actually they do and why ? bc they are powered and spin at a constant turning rate as long at they are powered.

Change in direction will be noticed ofcourse and the force created by the spinning gyro is 'translated' into a output signal. More change = more output.

I will not comment you any further as you have no idea how drones do function.

But all the best and many happy landings,
cheers
JJB

JJB* · 2021-10-26

mkh Posted at 10-26 08:28
As far as I remember the compass was pointing in the same direction. Also it usually complains and instructs to re-calibrate when it detects the anomaly during the start up.

In this case, there was no indication of the compass error.

so wait for DJI to give their assesment, as said in my posts more data in your log is not correct, vertical speed and baro height. So perhaps a IMU fault wich causes this behaviour of your drone.

cheers
JJB

Labroides · 2021-10-26

mkh Posted at 10-26 08:28
As far as I remember the compass was pointing in the same direction. Also it usually complains and instructs to re-calibrate when it detects the anomaly during the start up.

In this case, there was no indication of the compass error.

I suggest you read my earlier explanations, which tells you exactly what happened.

Labroides · 2021-10-26

JJB* Posted at 10-26 06:14
The only input to a gyro is its power supply, gyro gives an output signal it moves in its direction.
Sorry that you cannot follow my writing, this is how it works.

I think you are on your own with that idea.

Labroides · 2021-10-26

Johnny_J Posted at 10-26 05:37
Labroides wrote: "I went to some trouble to explain why and I'm not going to type it all out again"

I asked you what posts in this thread you were referring to.

I asked you what posts in this thread you were referring to.
There where two or three that I laboriously wrote to explain why the drone couldn't just hover or come back, just before I replied to you.
With your technical abilities, I didn't expect you to have any trouble going back a couple of posts and finding them.

JJB* · 2021-10-27

Labroides Posted at 10-26 12:33
I think you are on your own with that idea.

Hi Labroides,

Oke, no problem to be on my own (+ all the men and woman in the aircraft industries)

see 2 typical examples, read about the input and output....

cheers
JJB

Eheran · 2021-10-27

There is nothing spinning in those gyros here, its only a very small vibrating mass like the quartz in a clock. Read more about it here. Rotating gyros are for aircraft/boats, since its a much larger system. Those also dont suffer from the issues of the smaller MEMS devices like huge drift (output value is not 0 even if there is no change, at the same time its more on one side of 0 compared to the other). Which is also why the drone needs GPS as a reference, even tho GPS is way too slow for actually flying a drone. The drone could flip/crash/bla between two GPS values, even at 10 Hz. But GPS is long term stable, thats the big deal. Drift of the GPS signal is very limited. But you can still see the effect of bad GPS (and thus drift of the apparent GPS-position) while flying under a tree or what not: The drone drifts more.
The magnetic field is another reference, helping to reduce the rotational drift (where the drone is facing). Bringing a magnet close to the drone (no matter how well it matches the strenght of earths magentic field) should only cause more drift around its axis.

The actually flying and stabilization is done by the accelerometer and gyroscope (combined they are the INS, inertial navigation system), both having high rate outputs, something like 1000 Hz, suitable to correct the drones position quick enough even in high wind. Unlike GPS they only have short term stability due to the high drift. The only sensible way for such a fly-away to happen is when there is a issue with these signals. GPS/visual or magnetic sensing should not be able to cause more than bad drift, since thats exactly what they try to reduce. They cant cause such quick reactons of the drone. Flying from the outside into a no-GPS + false magnetic field situation (like a building or tunnel) would, otherwise, instantly and always cause a crash. This is not the case, but drift does indeed go up.

The drone combines all of these sensors (sensor fusion), using the short and long term stable inputs to get the best possible combined result.

DronePerspectiv · 2021-10-27

Seems exactly what I happened once. See my post https://forum.dji.com/forum.php?mod=viewthread&tid=234302
After that I never had this issue again.