Take off from boat - word of warning

[hydrolift]

Hi all!

Just had my first fly-away today which was scary. From reading other posts I assume I had magnetic interference when taking off.

I hand launched the MM from the aft/rear of a small boat with a 200hp outboard motor. I have launched the MM from the same boat the day before, but from the stem/front without any problems.

What happened was that the MM lanuched ok, but as soon as I gained altitude and made some distance from the outboard motor, it suddenly darted away forward att full speed. There was nothing I could do to stop it. I tried to turn e t c, but the MM just sailed away at full pace and I realized that I quickly had to do something drastic. So I pulled both sticks downwards until it crashed in a field of thorn bushes. We were boating in a canal at the moment, so luckily there was land to crash the MM in/on.

I think that the outboard motor generated a magnetic interference which affected the "magYaw" input. As soon as the MM was away from the outboard motor, it freaked out and became virtually uncontrollable.

It seems to be unharmed though. I think I only need to change the props. What bothers me a little bit is that the drone never recovered in flight. If, according to the GPS, the drone travels at speed without any input from the RC - why does it not stop flying away??? In my opinion there must be an overlooked bit of code there.

Lesson learned? Never launch the MM from near the outboard motor or other electronics (which many of you already know and so do I now by learning the hard way).

 

[Phantomrain.org]

Hi all!

Just had my first fly-away today which was scary. From reading other posts I assume I had magnetic interference when taking off.

I hand launched the MM from the aft/rear of a small boat with a 200hp outboard motor. I have launched the MM from the same boat the day before, but from the stem/front without any problems.

What happened was that the MM lanuched ok, but as soon as I gained altitude and made some distance from the outboard motor, it suddenly darted away forward att full speed. There was nothing I could do to stop it. I tried to turn e t c, but the MM just sailed away at full pace and I realized that I quickly had to do something drastic. So I pulled both sticks downwards until it crashed in a field of thorn bushes. We were boating in a canal at the moment, so luckily there was land to crash the MM in/on.

I think that the outboard motor generated a magnetic interference which affected the "magYaw" input. As soon as the MM was away from the outboard motor, it freaked out and became virtually uncontrollable.

It seems to be unharmed though. I think I only need to change the props. What bothers me a little bit is that the drone never recovered in flight. If, according to the GPS, the drone travels at speed without any input from the RC - why does it not stop flying away??? In my opinion there must be an overlooked bit of code there.

Lesson learned? Never launch the MM from near the outboard motor or other electronics (which many of you already know and so do I now by learning the hard way).

We have had two fly aways both from the boat and each one was do to Magnetic Interference causing Yaw Errors that made it impossible to stop it from flying away. DJI replaced both drones with %100 coupons.

Cars and Boats : Each boat and each car is different. The problem with the code for me is that when you have Magnetic Interference its a Flash on the screen that disappears before you can focus and since it goes away you think your ok but the reality is your Compass has been hit with a sledge Hammer .

I have demonstrated this on the Phantom 4 by using the Magnet inside the Iphone to induce atti mode .

 

[AMRE2ME2]

When you startup, you get that audio: “Homepoint is updated please CHECK IT ON THE MAP“
Isn’t this the point of that audio alert? Because if the drone is not pointing at the same heading as what is portrayed on the map, you are having interference!

Ive never had a flyaway with over 400 flights, but I do check that my heading and map are reading the same. I assumed this is why we get that reminder With every flight.

 

[Deleted member 103366]

If you post the logs here one of the 'guys' will probably delve into them and tell you what happened.

There is something called the "CSC" maneuver, relevant bit of the manual is top of page 39 etc.
You are supposed to be able to stop the motors in mid flight with it, it has two modes of operation (bottom of page 32), "emergency only" and "anytime".
IMO emergency only is a liablility, if the pilot thinks they need to stop the motors in mid flight and the mode is "emergency only" the command will not succeed.
As I now understand the manual, something dire has to have happened before "emergency only" mode will work and I suspect the drone would already be falling by the time the CSC was applied
Following some posts on the DJI forum I was tempted to try this and I have held the CSC position for up to 10 secs and the motors have not stopped. The drone descended, 'twisting' but as soon as the sticks were released it hovered perfectly normally.
I switched to anytime mode and tried it low, over shortish grass in tripod mode (slowest descent speed), the motors stopped after the position had been held for about 2 seconds, the drone fell maybe 18".
I would not suggest flying with the option set to "anytime" but it might be worth remembering that it is there. It is of course a last resort.

 

[ZaratustrA]

I would not suggest flying with the option set to "anytime" but it might be worth remembering that it is there. It is of course a last resort.

just my 2 cents: unlike the Mavic Pro you can't restart motors midair once stopped.

 

[Deleted member 103366]

I think that is debatable and depends on whether the mini tumbles, if it tumbles I think it will be impossible to restart the motors.
One of the software hackers might be able to confirm if there is a software switch that prevents a restart beyond certain angles of tilt, I am glad you raised the issue of restarts and brought the latter to mind, thank you.
That said, I would assume an attempt a mid air restart of the motors of an unhacked Mini will fail ...........But in this thread the objective was to get the drone down as quickly as possible so a mid air restart is not a consideration.

 

[Sooty]

I found that the outboard has no effect on launching or flying the drone. The problem I had was the depth sounder GPS and radio etc. Turning off the main switch to these electronics when flying solved all problems.

 

[boblui]

.... If, according to the GPS, the drone travels at speed without any input from the RC - why does it not stop flying away???

Because the drone was trying to correct for position drift due to wind. If the yaw sensor is off, it will never be able to get back to the original position. If the yaw error is relatively small, it will move in a circle. If the error is more than 90 degrees, it will fly away like what you have experienced. This is my Mini with the yaw sensor off by about 180 degrees :

 

[Meta4]

Lesson learned? Never launch the MM from near the outboard motor or other electronics (which many of you already know and so do I now by learning the hard way).

The description of the incident suggests that your guess about the cause was (almost) correct.
Something deflected the compass which the IMU uses as a reference to initialise the gyro sensors.
It might not have been the outboard motor itself and could have been much smaller and closer.

On a boat, launching from teh deck is problematic and hand launching and catching is preferred as a way of preventing issues.

What bothers me a little bit is that the drone never recovered in flight.
If, according to the GPS, the drone travels at speed without any input from the RC - why does it not stop flying away???

The drone is actually trying to correct for the problem it detects, but because the gyro sensors are providing incorrect data, each correction only makes it worse, which accounts for the acceleration you observed.

 

[mxz1998]

This is why you should check the map right after takeoff and make sure it shows that the drone is facing the same direction as the arrow shows it is.

 

[hydrolift]

I found that the outboard has no effect on launching or flying the drone. The problem I had was the depth sounder GPS and radio etc. Turning off the main switch to these electronics when flying solved all problems.

Ok, I was rather near the motor (approx 1m ) and about 2 m away from other electronics onboard the boat so I assumed it was the alternator from the outboard motor that caused the interference.

 

[hydrolift]

The description of the incident suggests that your guess about the cause was (almost) correct.
Something deflected the compass which the IMU uses as a reference to initialise the gyro sensors.
It might not have been the outboard motor itself and could have been much smaller and closer.

On a boat, launching from teh deck is problematic and hand launching and catching is preferred as a way of preventing issues.


The drone is actually trying to correct for the problem it detects, but because the gyro sensors are providing incorrect data, each correction only makes it worse, which accounts for the acceleration you observed.

I actually did hand launched the **** thing.

Just wished there was some sort of reset button on the RC for these types of incidents that would reset affected sensors to what they were set to the last flight for instance, if possible.

 

[hydrolift]

This is why you should check the map right after takeoff and make sure it shows that the drone is facing the same direction as the arrow shows it is.

Noted! Doubt that I would have had time to do anything in my case though. It all happened very quickly.

 

[Meta4]

The problem I had was the depth sounder GPS and radio etc. Turning off the main switch to these electronics when flying solved all problems.

They might have caused a little radio interference, but it's unlikely these would have had any effect on the compass unless you were launching from within 1-2 feet of the electronics.

 

[Meta4]

Just wished there was some sort of reset button on the RC for these types of incidents that would reset affected sensors to what they were set to the last flight for instance, if possible.

That's not possible.
The gyro's need to be told where north is when they are initialised.
They get that from the compass.

 

[hydrolift]

That's not possible.
The gyro's need to be told where north is when they are initialised.
They get that from the compass.

Wouldn't it be possible to make the Mini hover on the spot (using GPS)? Does it need to know which way i north for that?

 

[Meta4]

Wouldn't it be possible to make the Mini hover on the spot (using GPS)? Does it need to know which way i north for that?

It's not just north, it needs its gyros to have an accurate directional reference so that it can work properly and go in the proper direction.
Because when it moves and the GPS data is telling it that it's not where it should be, the drone tries to get back to where it should be.
But using false data from the gyros, each move to correct is wrong and puts it further off course.
The drone is unable to stop and hover.

 

[BudWalker]

This is why you should check the map right after takeoff and make sure it shows that the drone is facing the same direction as the arrow shows it is.

This check should be done before takeoff. After takeoff it's too late.

 

[hydrolift]

It's not just north, it needs its gyros to have an accurate directional reference so that it can work properly and go in the proper direction.
Because when it moves and the GPS data is telling it that it's not where it should be, the drone tries to get back to where it should be.
But using false data from the gyros, each move to correct is wrong and puts it further off course.
The drone is unable to stop and hover.

Ok, think I got it!

 

[BudWalker]

Just to clarify. There is no yaw sensor and the gyros don't directly determine yaw (or heading). The gyros can't be "reset". There is almost never a problem with the gyros or any of the other sensors.

Yaw is a value that's computed by the Flight Controller using several sensor inputs. At power up Yaw is set to the value determined by the magnetometers. When there are enough sats the FC determines the geoDeclination based on the location. The geoDeclination is then used to correct the Yaw so that it's based on true north instead of magnetic north. After this the FC is using mostly the rate gyros, accelerometers and GPS data to determine changes in Yaw, not absolute Yaw. It also uses magnetometer data to make the small corrections that occur due to inaccuracies in the rate gyros and accelerometers.

Beyond this classical fusion scheme the FC also incorporates info from the vision system. Occasionally, the FC will make abrupt changes to the Yaw value if it decides Yaw has been incorrect. It's a complicated business.

EDIT:
@hydrolift is it possible that the MM was powered up at a location that was geomagnetically distorted and then moved to, a possibly clean, launch location? Why haven't you posted the logs?