Take off from boat - word of warning

[Dave Maine]

Note that the drone uses all of the information available to hover or move - barometric altimeter, compass, inertial guidance system and gps. This information and the stick inputs, are all synthesized into flight instructions for the motors. If there is a large conflict between the various inputs the drone does the best it can, which may be catastrophic.

There is no easy way to disable one of the inputs, as there is no way to know which input is wrong.

 

[hydrolift]

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?

Hi!
I didn't post the log since I only wanted to warn others. I was pretty sure that it was magnetic interference that caused the fly away. But if you want to take a look I can post the log (I am not sure which file it is, but let's start with this one). Maybe it can shed some light and help others (and me) to fully understand what not to do.

It CAN be possible that I powered up the MM when I was in the stem and the launched it when I was in the aft, near the outboard motor. But I don't remember.

Thanks!
//Olof

PS. Will add video capture too shortly.

 

[boblui]

There is no yaw sensor .....

Very true ...... I just invented that name to avoid going into the sensor fusion thing which can be too complicated for many. Describing it as "compass error" will be technically wrong because the compass is likely to be correctly calibrated.

 

[BudWalker]

Very true ...... I just invented that name to avoid going into the sensor fusion thing which can be too complicated for many. Describing it as "compass error" will be technically wrong because the compass is likely to be correctly calibrated.

.. and to be fair. My description leaves out several aspects - some that I know about and some that I don't.

 

[hydrolift]

Here is the video of the fly away. Launch at about 00:16

 

[LtcdData]

I was going to say that this sounds like my experience with launching off the boat and the mini flys away but I realized it wasn’t the mini moving but the boat drifting. But it sounds like your experience is different.

 

[joeyfri13]

I also had a run away from the boat, I was able to regain some control and managed to steer it away from a cliff and back over the water where I was able to get it to go into RTH mode where I got it close then guided it the rest of the way back to the boat. The motor was not running but the trolling motor was running and operates on a WiFi signal so I was thinking that maybe it provided some interference causing the run away. The drone was hovering behind the boat for a while and there was a little bit of cross wind as well. How do I look at the logs?

 

[Meta4]

the trolling motor was running and operates on a WiFi signal so I was thinking that maybe it provided some interference causing the run away.

At worst, interference might swamp your control signal and initiate RTH, but it will never cause the drone to fly away.

How do I look at the logs?

Go to DJI Flight Log Viewer | Phantom Help
Follow the instructions there to upload your flight record from your phone or tablet.
That will give you a detailed report of the flight.
Come back and post a link to the report it gives you.

 

[Tiger64]

I am a little confused. Is it possible that the mm couldnt get a GPS lock because boat is moving? If it did have home point , that would have been behind boat so why would it zoom ahead of boat? It appears to be responding to inputs towards end of clip as it rotates and changes direction. Was there video signal loss so you were unable to see it was manouvering?
Sorry for all the questions. I have a friend with a new boat and plan on doing some pics for him but really dont want to drop my mm in the drink or have it fly away.
I figured we should be stationary to launch then get moving for shoot.
I know home point will need changing but fly manually most of the time.
Dont plan on getting too low. Have done some flying over swamp and river and no issues at 12m altitude. ( do i need to be higher over sea?)
Good tip about having electronics off.
Will be keeping an eye on this thread to see what outcome is.
Any other tips?

 

[Meta4]

Is it possible that the mm couldnt get a GPS lock because boat is moving?

No .. You can get GPS lock in your car at 70mph or in a jet plane at 600mph.
He had 15-18 sats the whole time.

If it did have home point , that would have been behind boat so why would it zoom ahead of boat?

Because it wasn't trying to return to home.
What it was doing is complicated but explained in earlier posts.

 

[hydrolift]

I am a little confused. Is it possible that the mm couldnt get a GPS lock because boat is moving? If it did have home point , that would have been behind boat so why would it zoom ahead of boat? It appears to be responding to inputs towards end of clip as it rotates and changes direction. Was there video signal loss so you were unable to see it was manouvering?
Sorry for all the questions. I have a friend with a new boat and plan on doing some pics for him but really dont want to drop my mm in the drink or have it fly away.
I figured we should be stationary to launch then get moving for shoot.
I know home point will need changing but fly manually most of the time.
Dont plan on getting too low. Have done some flying over swamp and river and no issues at 12m altitude. ( do i need to be higher over sea?)
Good tip about having electronics off.
Will be keeping an eye on this thread to see what outcome is.
Any other tips?

I saw the drone the whole time. It did respond to my inputs and I could rotate the drone, but when I tried to steer back to the boat it seemed as if the drone by itself had even more input to go FROM the boat if you understand what I mean. I managed to do a small sweeping turn and crash it into the bushes.
I think if you take the advice given here in this thread and stay away as far as possible from all motors and electronics (and verify that the compass i pointing in the correct direction) before taking off, you should be safe. With that said it is always a little sketchy to launch from boats. The homepoint is moving and if you crash you will likely crash into the sea...

 

[BudWalker]

Here is the video of the fly away. Launch at about 00:16

Viewing the video it appears that the Yaw value in the .txt log is correct. Kinda puzzling since the flight sure looks like what's expected from an incorrect Yaw at launch. If you can retrieve the .DAT it may contain some relevant info. It'll be FLY057.

 

[hydrolift]

Viewing the video it appears that the Yaw value in the .txt log is correct. Kinda puzzling since the flight sure looks like what's expected from an incorrect Yaw at launch. If you can retrieve the .DAT it may contain some relevant info. It'll be FLY057.

Ok, thank you for helping out! Here is the dat-file.

 

[AboriginalBrotherhood1162]

At worst, interference might swamp your control signal and initiate RTH, but it will never cause the drone to fly away.

Go to DJI Flight Log Viewer | Phantom Help
Follow the instructions there to upload your flight record from your phone or tablet.
That will give you a detailed report of the flight.
Come back and post a link to the report it gives you.

*Off the subject*Can any member post a flight record to learn the details of an unexpected flight glitch so that precautions could be taken to avoid a similar situation. Just curious how these forums work.

 

[Meta4]

*Off the subject*Can any member post a flight record to learn the details of an unexpected flight glitch so that precautions could be taken to avoid a similar situation. Just curious how these forums work.

An unexpected flight glitch could be anything and there would be many different strategies.
Probably the best help for you would be to suggest reading posts in this sub-forum: Mavic Crash & Flyaway Assistance
That gives you the valuable opportunity to learn from the hard earned lessons and misfortune of others.

 

[boblui]

Ok, thank you for helping out! Here is the dat-file.

This is a very weird case. Immediately after take off, the direction sensed by the IMU was still about correct. There was no significant magnetic interference. Then rudder was apply and the craft turned by about 180 degrees as shown in the footage but the IMU outputs indicated that the craft has turned by about 360 degrees. It seems that the direction sensors suddenly went bad during that turn. I will leave it to more experienced analysts to investigate further.

 

[hydrolift]

This is a very weird case. Immediately after take off, the direction sensed by the IMU was still about correct. There was no significant magnetic interference. Then rudder was apply and the craft turned by about 180 degrees as shown in the footage but the IMU outputs indicated that the craft has turned by about 360 degrees. It seems that the direction sensors suddenly went bad during that turn. I will leave it to more experienced analysts to investigate further.

Thank you for helping out!

Yes, the MM hovered ok over the boat until I turned it 180 degrees, in the direction of the boat. From there things got weird and I got very surprised that it suddenly took off like that.

My first assumption was that as soon as it left the magnetic interference area of the boat, the sensor's input changed. But of course, I could be completely wrong.

 

[hydrolift]

Another thing that might be of help to understand the behaviour... At 2 other occasions when I tried to land the MM in my hand it slowly twisted clockwise at the same time as the landing sequence started and it started to descend. Caused bloodshed...

I don't know if this is common or if it has anything to do with the incident on the boat.

 

[Meta4]

My first assumption was that as soon as it left the magnetic interference area of the boat, the sensor's input changed. But of course, I could be completely wrong.

Yaw errors usually don't show up until the drone moves out of the magnetic field and turns enough, and then there's a difference between gyro data and gps/compass data.

 

[BudWalker]

This is a very weird case. Immediately after take off, the direction sensed by the IMU was still about correct. There was no significant magnetic interference. Then rudder was apply and the craft turned by about 180 degrees as shown in the footage but the IMU outputs indicated that the craft has turned by about 360 degrees. It seems that the direction sensors suddenly went bad during that turn. I will leave it to more experienced analysts to investigate further.

Weird indeed. This is one of those cases where the FC made an abrupt change to the Yaw value. At 8.411 secs Yaw abruptly changes from -149° to 41°. But, both the magnetometers and Z axiz gyro don't record that change. Erratic flight then ensued.


Why did the FC (erroneously) decide that Yaw was incorrect and needed fixing? Still thinking about this.

Maybe @sar104 has something to add.