DJI Air 2s uncontrollable flight and crash.

[Massabusa]

LOG- DJI Flight Log Viewer - PhantomHelp.com

Normal flight good conditions- until the drone all of a sudden took off (straight ahead at full speed) without any input from me. I tried many times to get the RTH to function, and you can see from the log it returned to directly above home several times. The RC would intermittently ask if I wanted to cancel the landing/RTH- again without input from me and the RC. It finally crashed into a tree 1000 feet away from me, out of site- behind another house where I used "Find my Drone" to locate it. Some damaged props, a few bruises to the exterior. I got home and swapped out all 4 props, and tried to initiate a short test flight to see if the drone was operational. I took off and hovered at 3 feet, found that I had no control with the right stick for forward and backwards flight (mode 2 in use). I landed it and tried to calibrate the RC, was not able to get any signal from the right stick, so the calibration wouldn't complete. I tried many times and did get it to calibrate, but on the next quick flight = again hover at 3 feet, no forward or backward flight. I tried another Iphone, same result, wouldn't complete the RC calibrate due to no right stick. Looking at the log, it looks like it thought there was a full forward right stick, as it flew off at 30MPH... I've sent the RC and the drone off to DJI for service.
Anybody else seen this failure?
Is there anything else I could have done?
I was kind of in a panic.

 

[vindibona1]

Your method of testing seems clear and it appears that there is something wrong with the controller or how the drone is receiving (or not) and how it is not responding to commands from the right stick.

I can only imagine how disconcerting that was. I experienced something like that with my first cheap Chinese drone that seemingly arrived defective and took off on it's own and crashed on day one. Ultimately I got refunded by my bank (not Banggood who lies about defects and returns on their website). I used the refunded money to buy my first decent drone, a Mini 1.

It will be interesting to see if DJI reports back to you on the issue or just sends you a new drone without comment or information.

 

[Yorkshire_Pud]

Phew that was a tough one.
The following comes with the benefit of hindsight and having thought about your flight, I doubt I would have thought of it if I had been in your shoes at that time.

As you say the drone seems to have 'thought' that the controller was giving a combined full left and a full forward command from around 303sec onwards and acted accordingly.

The manual states that during RTH (which I take to mean the climb to RTH height and the actual flight towards home)
" The orientation of the aircraft and the direction of flight, however, cannot be controlled."

That seems to fit your log. From 353sec to 364sec the drone is climbing to RTH height and does not move horizontally. At 364sec you cancel the RTH and it 'takes off' again. At 384sec you initiate RTH and the drone starts to 'behave' and heads for home. At 418sec it arrives over the home point and switches from GoHome to Autoland.
As soon as it does that it 'takes off' again, which is what one wants during landing i.e. one wants the controller to be able to move the drone during a landing just incase the would-be landing spot needs adjustment.

With the exception of 537sec to 544sec (no idea what happened in there) this pattern is repeated for the remainder of the flight.

What could you have done?
You could have switched the controller off.
Counter intuitive I know but, assuming it is the controller that has gone wonky, that would have stopped any commands being sent to the drone and the drone would have failsafe RTH'ed. There would have been no unwanted commands during the Autoland phase and the drone would descended vertically and landed.

Other than that, I think you would have had a very difficult to impossible task to land the drone safely during any descent. If you had given it yaw, to keep it pointed at home, it would have been descending in a helix.

You could have possibly reduced the horizontal speed by switching to cine mode but you would still have had the helical landing phase.

Again, all the above is in hindsight and I very much doubt I would have thought of it in the heat of the moment.

 

[Meta4]

Anybody else seen this failure?

Variations of this have frequently been reported.

Do you remember which direction your drone was pointing when it was on the ground at launch time?
The data indicates that it was just a little east of north, but I think this is not correct.
Can you describe the launch spot?
What was the surface you launched from?

 

[Massabusa]

Variations of this have frequently been reported.

Do you remember which direction your drone was pointing when it was on the ground at launch time?
The data indicates that it was just a little east of north, but I think this is not correct.
Can you describe the launch spot?
What was the surface you launched from?

It was in fact almost north at launch.
Launch area is a small back yard with larger trees and a house surrounding.
Grass/lawn.

 

[Massabusa]

Phew that was a tough one.
The following comes with the benefit of hindsight and having thought about your flight, I doubt I would have thought of it if I had been in your shoes at that time.

As you say the drone seems to have 'thought' that the controller was giving a combined full left and a full forward command from around 303sec onwards and acted accordingly.

The manual states that during RTH (which I take to mean the climb to RTH height and the actual flight towards home)
" The orientation of the aircraft and the direction of flight, however, cannot be controlled."

That seems to fit your log. From 353sec to 364sec the drone is climbing to RTH height and does not move horizontally. At 364sec you cancel the RTH and it 'takes off' again. At 384sec you initiate RTH and the drone starts to 'behave' and heads for home. At 418sec it arrives over the home point and switches from GoHome to Autoland.
As soon as it does that it 'takes off' again, which is what one wants during landing i.e. one wants the controller to be able to move the drone during a landing just incase the would-be landing spot needs adjustment.

With the exception of 537sec to 544sec (no idea what happened in there) this pattern is repeated for the remainder of the flight.

What could you have done?
You could have switched the controller off.
Counter intuitive I know but assuming it is the controller that has gone wonky that would have stopped any commands being sent to the drone and the drone would have failsafe RTH'ed. There would have been no unwanted commands during the Autoland phase and the drone would descended vertically and landed.

Other than that, I think you would have had a very difficult to impossible task to land the drone safely during any descent. If you had given it yaw, to keep it pointed at home, it would have been descending in a helix.

You could have possibly reduced the horizontal speed by switching to cine mode but you would still have had the helical landing phase.

Again, all the above is in hindsight and I very much doubt I would have thought of it in the heat of the moment.

Thank you for looking and your reply. I never cancelled the RTH, I think the drone interpreted the “phantom full right stick forward” signal as the operator overriding the RTH? I had no idea what was happening and never even considered the idea of a controller malfunction for some reason. I have only about 3-4 weeks of experience, and a dozen short flights at this point.

 

[Capt KO]

Sounds like a warranty case anyhow.

 

[Massabusa]

Sounds like a warranty case anyhow.

I hope DJI agrees with you.

 

[Meta4]

It was in fact almost north at launch.

I was in a hurry earlier and it turns out not to be what I first thought.
The cause is simple now that I have time to look at the data.
At 5:03.2 your problem started.
From then until the end of data, your right joystick was pulled fully left and pushed fully forward or there was a fault in the controller to give that input to the drone.
With the drone trying to fly at full speed left/forward, everything else was crazy.

It's most likely that there was nothing wrong with the drone and the controller was the whole problem.

 

[Massabusa]

I was in a hurry earlier and it turns out not to be what I first thought.
The cause is simple now that I have time to look at the data.
At 5:03.2 your problem started.
From then until the end of data, your right joystick was pulled fully left and pushed fully forward or there was a fault in the controller to give that input to the drone.
With the drone trying to fly at full speed left/forward, everything else was crazy.

It's most likely that there was nothing wrong with the drone and the controller was the whole problem.

I agree. Thanks for taking a look at the log!
Controller fault, but I hope DJI repairs or replaces the drone after their inspection. It hit several branches a few minutes before the crash, and hit branches hard in the final crash that had the drone spiraling down to the grass back yard from a good height with 2 broken props..

 

[Yorkshire_Pud]

I never cancelled the RTH, I think the drone interpreted the “phantom full right stick forward” signal as the operator overriding the RTH?

Whether, at 364sec, you cancelled the RTH or whether it was some automated process that cancelled it I do not know but at 364.1sec the csv of the log contains an entry "Return to home cancelled".

If Meta4 or @sar104 or @slup etc. can come up with an answer I would be interested in knowing why the drone flew more or less straight between 537.7sec & 544.7sec whilst in "Gentle GPS". Gentle GPS seems to be connected to APAS being enabled.

 

[TheRock]

I am still working on getting the info from the Smart Controller, but same thing happened to me. The drone just went nuts and spiraled into the sea. Had a perfectly normal 18 minute flight just 5 minutes before this. Brought it home. Put in new battery. I did notice the drone had trouble pairing with the SC (took 3 tries). Took off fine, got it to a spot where I wanted to start recording video. Hit the video button and it just went nuts, spiraling out of control then down. When I hit the sticks for up, no response. Now sitting in beautiful Nikko Bay. Will try to get flight records for you soon. But everything was updated and seemingly fine. I have been flying for five years and had this drone since the Air 2S came out and never had a problem in many flights. So strange. Must be an IMU problem of some sort I am guessing. Seems odd that just out of the blue the drone goes bezerk. There's nothing around here at all that would cause interference.

 

[Meta4]

I am still working on getting the info from the Smart Controller, but same thing happened to me. The drone just went nuts and spiraled into the sea.
.... Hit the video button and it just went nuts, spiraling out of control then down.

The data is always the place to go to find details.
But if your drone actually spiralled down, that suggests strongly that it lost a prop or a motor, which wouldn't be the same thing as the OP's case.

Must be an IMU problem of some sort I am guessing. Seems odd that just out of the blue the drone goes bezerk.

Yeah ... that's just a guess, and an unlikely one.

There's nothing around here at all that would cause interference.

Interference won't ever cause crazy, unpredictable flight.
At worst, it might swamp your control signal and initiate RTH.

Find your flight data and start a new thread with it.
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.
Or .. just post the txt file here.

 

[TheRock]

The data is always the place to go to find details.
But if your drone actually spiralled down, that suggests strongly that it lost a prop or a motor, which wouldn't be the same thing as the OP's case.

Yeah ... that's just a guess, and an unlikely one.

Interference won't ever cause crazy, unpredictable flight.
At worst, it might swamp your control signal and initiate RTH.

Find your flight data and start a new thread with it.
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.
Or .. just post the txt file here.

Thanks, I read that link but my MAC doesn't seem to recognize my SC. Can't locate the file. (yet) Prop loss is unlikely. It did a couple of odd things the day before which is why I made sure everything was updated before these 2 flights but nothing like this. I'll get the info and start a new thread as you suggest. Thank you.

 

[Meta4]

I never cancelled the RTH, I think the drone interpreted the “phantom full right stick forward” signal as the operator overriding the RTH?

Joystick input doesn't cancel RTH.

 

[Meta4]

Whether, at 364sec, you cancelled the RTH or whether it was some automated process that cancelled it I do not know but at 364.1sec the csv of the log contains an entry "Return to home cancelled".

RTH was cancelled twice - at at 5:08.3 and again at 6:04.1.
But the data only shows the RTH button being pressed one, at 5:08.6.

If Meta4 or @sar104 or @slup etc. can come up with an answer I would be interested in knowing why the drone flew more or less straight between 537.7sec & 544.7sec whilst in "Gentle GPS". Gentle GPS seems to be connected to APAS being enabled.

It's difficult to follow everything that's showing in the data.
The straight-line flight is due to the spurious joystick inputs, but the cause of abrupt changes of direction isn't shown.
Between 537.7sec & 544.7sec, as well as the OSD.flycState showing GentleGPS, the OSD.flycCommand is showing Dynamic Home Point Open.
I've not seen that before and don't know what it means.

 

[Yorkshire_Pud]

, but the cause of abrupt changes of direction isn't shown.

Maybe I am missing the point you are trying to make but don't they correspond to the start of successive RTH's or Go-Home's or the early part of them?
I see this behaviour with my Minis and M2P/Z as the drone points itself towards home.

I had prepared this earlier but it didn't seem necessary but it does show what I 'see'.

 

[Yorkshire_Pud]

OSD.flycCommand is showing Dynamic Home Point Open.

I am not sure if "I've not seen that before and don't know what it means" relates to "Dynamic Home Point" but if not, does that mean the Air 2s have has automatic [edit updating] of the home point if the pilot and controller go for a wander?

I have seen "Dynamic Home Point" or something similar in either the logs or App of one of my drones but can't remember which drone.

 

[Meta4]

Maybe I am missing the point you are trying to make but don't they correspond to the start of successive RTH's or Go-Home's or the early part of them?
I see this behaviour with my Minis and M2P/Z as the drone points itself towards home.

If they were going towards home, they would be pointing towards home ... but they are all over the place.

 

[Meta4]

I am not sure if "I've not seen that before and don't know what it means" relates to "Dynamic Home Point" but if not, does that mean the Air 2s have automatic of the home point if the pilot and controller go for a wander?

I have seen "Dynamic Home Point" or something similar in either the logs or App of one of my drones but can't remember which drone.

I know what dynamic home point is, but I don't know why this has showed up in the data or what it means there.
There are few odd things happening in the data ... it's not as simple and clear-cut as most are.