Mini 2 tried to fly away on me.

[teeroy]

A friend had bought some sea cans for storage units and asked if I could fly up above his yard and take a few photos to check that they were set up nice and straight in relation to his property lines and house. There was some wind but I didn't think it was severe.

I started the Mini 2 and let it hover for a bit, then went up to about 60 meters height. Without warning of strong wind or the likes from the controller, the AC started to drift away at a pretty good pace to the east with no input on the sticks. I immediately tried to fly it back but it just kept going so I gave it full down on the left stick while continuing to hold it in "reverse". It started dropping altitude but kept going east away from me as it came down. I was approaching a bluff of trees and as it got to the point where I had to decide to let it stay up and keep going to the field on the other side of the trees or keep dropping and crash into them, all of a sudden it stopped its easterly movement and came straight down just in front of the trees. I did get a wind warning at one point, but not before it started heading east.

After I got it under control and flew it back to my original point, I tried to go up again a little further west in case it did it again and I knew I had the space to bring it down before those trees. I flew up to as high as 120 meters with no further issues.

Was this a potential blow away? Any help or criticism would be appreciated. If it was my fault I will take my lumps, I just want to know what happened here. Let me have it.

DJI Flight Log Viewer | Phantom Help

Upload and view flight logs from your DJI drones

www.phantomhelp.com

 

[teeroy]

Couple shots I got after it seemed ok.

 

[teeroy]

here are the DAT and TXT files if somebody wants to plug them into something else

 

[Deleted member 103366]

Ooooo I wonder if this means the wizards have cracked the Mini2 encyrption problem, a cursory glance at the DJIFlightRecord_2021-01-13_[12-59-58][1]-TxtLogToCsv.csv downloaded from the Phantomhelp page shows it to be full of data

For the moment and until one of the wizards come in please do not update the APP.

 

[Deleted member 103366]

What I think is the elevator/pitch record in that csv, column CM / 91 / "RC.Elevator". seems to agree with the RH stick indicator on Phantomhelp and the drone did not intially respond to it.

 

[teeroy]

What I think is the elevator/pitch record in that csv, column CM / 91 / "RC.Elevator". seems to agree with the RH stick indicator on Phantomhelp and the drone did not intially respond to it.

It just kept going, but was it blowing away or just on it's own little walkabout? I sure didn't tell it to go east, and it ignored me when I wanted it to come back!

 

[Deleted member 103366]

Yeah, if you look at the attitude indicator on the Phantomhelp page (the blue semi circle thing) it was almost level during that nasty drift, later where you have control it pitches etc. in response to R stick inputs, (the blue section of the inidcator moves up and down and tilts etc.
It will be interesting to see what the wizards say.

 

[teeroy]

Yeah, if you look at the attitude indicator on the Phantomhelp page (the blue semi circle thing) it was almost level during that nasty drift, later where you have control it pitches etc. in response to R stick inputs, (the blue section of the inidcator moves up and down and tilts etc.
It will be interesting to see what the wizards say.

Hey, thanks for that. I didn't know what the blue thing in the circle meant. That certainly doesn't give me a warm feeling inside. If it was wind it should have been at max tilt trying to fight the wind...hmmm.

 

[boblui]

At the height of 190 ft where the drift occurred, there was a rather strong westerly wind as can be seen at 2min 23sec when the control over the craft has resumed. The craft need to tilt forward ( nose down ) by 18 degrees in order to stay in place. I would say the wind speed was around 30 km/hr



From 30 ~ 56 sec, the pitch angle stayed at around 9 deg ( nose up ) which was not sufficient for countering the wind ( 18 deg needed ) so the craft drifted to the east. During that time period, it did not respond to full backward elevator input but it did respond to throttle-down input. No apparent reasons can be seen from the log. I suspect that it was a software glitch



After 56 sec. It resumed responding to all control inputs normally.

I hope other members can shed more light onto this strange case which I have never seen before.

 

[teeroy]

At the height of 190 ft where the drift occurred, there was a rather strong westerly wind as can be seen at 2min 23sec when the control over the craft has resumed. The craft need to tilt forward ( nose down ) by 18 degrees in order to stay in place. I would say the wind speed was around 30 km/hr

View attachment 121685

From 30 ~ 56 sec, the pitch angle stayed at around 9 deg ( nose up ) which was not sufficient for countering the wind ( 18 deg needed ) so the craft drifted to the east. It did not respond to full backward elevator input but it did respond to throttle-down input. No apparent reasons can be seen from the log.
View attachment 121687
View attachment 121688

After 56 sec. It resumed responding to all control inputs normally.

I hope other members can shed more light onto this strange case which I have never seen before.

Thank you sir, I appreciate your input. I'm just happy it smartened up and I didn't have to go searching the fields for it. But I'd sure like to know why it behaved that way.

 

[DanMan32]

The MM2 is supposed to be capable of 25° angle in normal mode and resort to 40° if it detected wind. That's in the AC specs on the DJI website.

Even the MM1 will tilt 20° in normal mode, 30° in sport but no high wind tilt override.

 

[slup]

Mmm ... this will be hard to solve without access to the raw sensor data from the DAT log (this as both the AC DAT & mobile device DAT is encrypted for the Mini 2).

But this is now the second case I see where the AC is flying in one direction but the tilt direction is indicated to be in the opposite direction. This may very well explain why it seemed like the AC didn't listen to the negative elevator stick input ... & flew away uncommanded.

This kept on until the AC was turned around to be flown back ... stuck gyro?



The same shown in chart form below ... green the yaw & flying direction ... & blue tilt direction. nearly 180 degrees difference.

At 68sec into the flight they come in sync for the first time in the flight ... it's then the AC becomes responsive again.



It was rather windy above 60m ... but not so much that the Mini 2 should drift & not manage to hold position, (which should be the only case where tilt direction vs. flying direction could behave like this).

 

[Deleted member 103366]

To Boblui, Slup and Budwalker etc., do you think that the tilt angles depicted in/on the Phantomhelp page (OSD.pitch & OSD.roll)? reflect the actual tilts of the drone
If the thinking that this might be down to a stuck or sticky sensor or gyro would a bit of 'acrobatics' at a low height serve to test for it?
If a "software" glitch is it thought that it would be entirely random in its appearing and disappearing?

 

[slup]

To Boblui, Slup and Budwalker etc., do you think that the tilt angles depicted in/on the Phantomhelp page (OSD.pitch & OSD.roll)? reflect the actual tilts of the drone
If the thinking that this might be down to a stuck or sticky sensor or gyro would a bit of 'acrobatics' at a low height serve to test for it?
If a "software" glitch is it thought that it would be entirely random in its appearing and disappearing?

Can't really say without the raw sensor data available ... & we know that gyros can get stuck & that computational error have happened before but they are really rare, it's usually depending on sensors reporting wrong. But acrobatics ... nah, if you don't by that mean poking it into a solid object a couple times

 

[Deleted member 103366]

Ta, by acrobatics I meant a couple of gentle or sharp, forwards, backwards, sideways, commands etc. just to see if it behaved itself.

 

[boblui]

To Boblui, Slup and Budwalker etc., do you think that the tilt angles depicted in/on the Phantomhelp page (OSD.pitch & OSD.roll)? reflect the actual tilts of the drone
If the thinking that this might be down to a stuck or sticky sensor or gyro would a bit of 'acrobatics' at a low height serve to test for it?
If a "software" glitch is it thought that it would be entirely random in its appearing and disappearing?

I am not sure if the sensors were OK but the fused sensor outputs ( green path ) and GPS ( red path ) clearly told the flight controller that the craft was drifting but the actual movement of the craft suggests that the flight controller did nothing to stop the motion and ignored elevator control input until it waked up later. The behavior of the flight controller was definitely abnormal.



There is a recent case ( Flight logs can someone check them ) in which the accelerometer output was biased hence a bias in the indicated pitch angle but the craft was still able to fly normally and hold position. Only the maximum speed was reduced and false strong-wind warning was issued.

If it's me, I will not be comfortable to fly it again unless DJI can come up with some assuring explanations.

 

[BudWalker]

I also think the sensors are OK. As observed by @boblui above the flight behavior is consistent with those data.

The gyros are rate data; i.e. a "stuck" gyro would manifest as constant rotation about the axis.

The simplest explanation, i.e. requiring the weakest assumptions, is that the FC commanded insufficient tilt. Too bad we can't see the .DAT

 

[teeroy]

Ta, by acrobatics I meant a couple of gentle or sharp, forwards, backwards, sideways, commands etc. just to see if it behaved itself.

I did exactly that, but inside my house. When I got home there was an AC software (firmware?) update message that came up so I did the update and then fired it up once it had good satellite acquisition. For some reason I get good coverage inside the house. I let it hover for about 5 minutes and it stayed rock solid with no drift. So then I did some fast ahead, reverse, and side to side and abrupt stops, it performed flawlessly.

I am a little apprehensive about flying it outside for now, how would I go about letting DJI know there was a glitch, like a warranty claim?

Thanks to you fellas for examining the logs and giving your input. It is certainly appreciated. I respect and honor your knowledge.

 

[Deleted member 103366]

Umm one point, and a good one I didn't think of until now, low down it can use the downward looking sensors to hold position and I think they provide better location than GPS.
How they interact with what ever caused your fly away I do not know, sorry, and as a 'bummer' their range might be 30m, ho hum, complicated machines these toys are!

 

[teeroy]

YOUR MINI 2 WILL DO THIS TOO.

This attempted flyaway issue by my Mini 2 was solved to be a programming error in the AC firmware. Evidenced in this thread...

Mini 2 has the pitch angle limited when the camera is pointing straight down

Just wondering if any owners of the craft has experienced it. This problem has been reported by two local pilots and this log is provided by one of them. Near the end of the flight the camera was pointing straight down and the drone was unable to counter the moderate wind and the speed was...

mavicpilots.com