I have been thinking a lot about what happened during my I1 "flyaway" a couple of weeks ago, but it was not until this evening that I figured it out! I can conclude that my I1 did not appear to by definition experience a "flyaway." I finally obtained access to the flight log and the moment I reviewed the details of this flight, it was clear what happened. (sorry, I can not share video/logs due to an NDA on this project), but I assure anyone that I have nothing to gain from posting the findings of our investigation in to this incident.
So, here is what happened:
On the previous 2 flights, it is clear that the craft/compass direction is facing away from my launch location, however, on the third flight, the compass orientation is flipped 180 degrees at takeoff, so the I1 thought it was looking at me, not away from me. This means that although the craft is visually orientated normally, the flight controls were all reversed.


This also explains why I was unable to regain control when I switched to atti, as this only accellerated the rate at which I was compensating in the wrong direction on all inputs, as I was responding according to the visual orientation of the craft. I also assumed that RTH did not engage, but maybe it did, just a moment too late as Atti mode had already gained too much momentum for GPS to respond in time before impact.

What does this mean? We have added a new step to our preflight checklist that should ensure this never happens again. I have never included looking at the orientation bubble to determine preflight craft orientation, as this is not a feature present on any UAS that I have operated. I have always calibrated the compass before my first flight and gone from there. Without this experience, I would have looked at the craft/gimbal orientation feature and assumed that the craft needed to be in the air and get better GPS lock before the orientation corrected itself, and would have taken off. This incident has taught me that the craft/gimbal orientation feature is actually correct and this should definitely be a preflight check.

I wonder if any of the other "flyaways" would show that this happened. I have read other posts where people mention reversed controls, so maybe this is some kind of bug, I honestly do not know, but I strongly believe this is what happened in my case and it appears that at least my incident could have been avoided if I had looked for and trusted the orientation information on the screen.

If this is a bug, I hope DJI is doing something to fix it. In all of my years flying, I have never experienced a sudden full reversal of all controls, so this is certainly concerning, but looking at the facts, I am certain my situation could have been detected/avoided.

I hope this helps and saves someone from a bad day, I surely learned something from it.

Thank you for sharing your discovery and findings. Can you post a screen shot on the app, showing us what you are looking at?

Rich

Hi Rich,
I do not know how to share images, but I will try to explain. In the lower left hand corner of the app, you will see a widget that shows the orientation of the craft in the form of a red arrow, and it shows the orientation of the gimbal in the form of a reversed green arrow/beam. In my case, my previous two flights showed the craft correctly orientated on the first two takeoffs/flights, but on the third (crash) flight, this widget showed the craft to be reversed at takeoff, meaning according to the flight controller, the red arrow and gimbal were pointing towards me at takeoff, instead of away from me. I hope this helps.

This make sense

I always check the orientation before take off - And I do of habit from flying my Phantom with a third party app. The app randomly show the phantom in wrong orientation and it  is very annoying. ( I have stopped using that app for a while ago )
I think its a very good point you have about checking the orientation before takeoff.

Good find.

Wow.

So if THAT'S the case, I wonder if it explains ALL the flyaways.  Consider: if the compass and therefore the controls are REVERSED, that would also fool the aircraft's internal logic - the logic that is used for RTH.  Once RTH is initiated it would accelerate the aircraft in the opposite direction from the intended one.  Unless there was a built in algorithm to trap for this condition and reverse the controls, it would be unrecoverable.  Even the attempt to hover in a light wind would result in a flyaway similar to that early crash video in which the aircraft instantly Rams into a garage door on takeoff, and correct stock inputs from the pilot would accelerate even faster the wrong way.

This could explain A LOT.

Did you calibrate the compass before the third flight? Any suggestions why the orientation of the inspire was changed?

I wonder if this is what happened to me.... I wish I had a video of my screen to really determine if that was my case. you can see my flight record... http://forum.dji.com/thread-14137-1-1.html

JKT - I just looked at your thread and have one question which may help determine if this is also what happened in your case. On the map, it shows that you took off facing the road at about a 45 degree angle (red arrow/green beam facing the road). Were you in fact facing this direction at takeoff?

StefT - Of course I calibrate prior to flight. I assure you, we operate under strict preflight guidelines. I have no idea why this happened. There clearly was an issue in the software. The flight computer reversed all inputs exactly 180 degrees, which I would assume either the craft reversed everything, or the remote failed and reversed inputs. I have no way of figuring out exactly what happened.

As soon as I pop up with any quad I do a little input test to make sure everything looks to be working... Spending the first 30 sec to a min to just shake out any bugs before flying any distance away is never a bad idea.

Stick input is direct without the compass influencing it unless its in IOC mode, only way how it would reverse the stick controls would be if it suddenly assumed that the flight controller is mounted 180 degrees rotated inside the craft

PanamonCreel - Thanks for your input. Can you offer any insight as to what would cause the craft to think the flight controller is suddenly mounted 180 degrees in the opposite direction?

info@thepixelbu Posted at 2015-5-29 01:02
PanamonCreel - Thanks for your input. Can you offer any insight as to what would cause the craft to  ...

negative, i have no insights on that and can only guess at buggy code,  lack of plausibility checks,  controller freeze up, controller failure.

Well, I just called the DJI LA office to report my findings and it seems they straight up do not believe me that this is what happened. They have "never heard of this issue" and will document it on my account, but will not take further action as my case is only a single one, which has never been documented before. It seems they felt my issue was one of incompetence, more than a failure of technology. So, I guess we should only report a problem if it is a known or common problem...great policy.

Try to run the simulator..i have my compass orentation reverse as well..running on the latest app...kindda anoying

PeteGould Posted at 2015-5-28 21:18
Wow.

So if THAT'S the case, I wonder if it explains ALL the flyaways.  Consider: if the compass and ...

Pete,

I seen something similar in the simulator.  See orientation of Aircraft in left and right corners

http://s1123.photobucket.com/user/GB114/library/Inspire1%20-%2001

You will see that aircraft is pointing in different directions.

Just to clarify, the displayed orientation was incorrect on the google map in the bottom right hand corner.

Can anyone else, who has experienced a loss of control, comment if your log shows your orientation to have been incorrect as well?

PeteGould Posted at 2015-5-28 21:18
Wow.

So if THAT'S the case, I wonder if it explains ALL the flyaways.  Consider: if the compass and ...

I have actually experienced this (all stick input was totaly messed up) ... In the simulator ...
But restarted the simulator and all was back to normal!

w1der Posted at 2015-5-29 03:54
I have actually experienced this (all stick input was totaly messed up) ... In the simulator ...  ...

Now if we could just put in another quarter after a crash, and restart the game, when actually flying...  

I have checked with my friend's Inspire and he also had the same issue...the compass orentation did not match the aircraft orentation...hope DJI can clarify on this as we both running the latest DJI app.

GB44 Posted at 2015-5-29 03:45
Pete,

I seen something similar in the simulator.  See orientation of Aircraft in left and right c ...

Scary stuff.  I wonder why it's happening to some but not others... must make it awfully difficult for the DJI folks to reproduce...

W1der - I wonder what would have happened if you had been flying the one time when your controls were messed up in the simulator. I assume that as the simulator is attached to the remote and craft, the actions of the simulator were accurate. This is just an assumption, but you may have been really luck there.

PeteGould - I agree, this scenario may be rather difficult to reproduce, but there are already several people in this thread that have confirmed they experienced the same.

firdausyusof198 - Thanks for your input. I hope that this is a/the real issue. I would be happy to know that this is a/the problem, as it can be detected/avoided rather easily before it becomes an issue. Now if only DJI would give us some information as to what is actually going on here! I know their goal is to separate themselves from the masses (us, the customer), and I know that us lowly end users could not possibly be  smart enough to comprehend the extent of their genius, but a little information would really be appreciated one of these days, especially as so many of us have lost and/or continue to lose confidence in this brand.

info@thepixelbu Posted at 2015-5-29 06:53
W1der - I wonder what would have happened if you had been flying the one time when your controls wer ...

Yes ... This thought crossed my mind also!

I know I where toggling between all 3 flight modes pretty aggresively at the time it happened just to learn the differens between them ... but as I remember the error did appear at take off, after landing and setting a new home point ... and the simulator do crash from time to time ...

No mather what the equipment should be able to handle this.

PeteGould Posted at 2015-5-29 04:53
Scary stuff.  I wonder why it's happening to some but not others... must make it awfully difficult ...

It is a worry.

I asked DJI why I the simulator showed this but got no response or feedback.

It seems like if everything was 180 degrees out of wack you would have noticed this at liftoff.??

kelleyre@gmail. Posted at 2015-5-29 23:55
It seems like if everything was 180 degrees out of wack you would have noticed this at liftoff.??

Agreed.
There is no way this could have been like this from the get go because someone would have noticed right away. Maybe this happened midway through flight?

thanks for sharing your info with us. i will personally be checking that on the map from now on before each flight i havn't been giving it much thought as it usually look about right. it may be a more critical reading then i gave it credit for.

kelleyre and sbarryjackson - I agree with both of you. The problem was not just that controls were reversed, and they were definitely not reversed right when I took off, however, when things went wrong, the controls did reverse, but I can not determine exactly when it happened or on which end it happened (ground or air).

I can add that I knew something was wrong right before completely losing control because the I1 started to drift hard right while I was hovering about 8 seconds after takeoff. I tried to correct this action with inputs and change of flight mode, and that is when I realized the I1 was not responding correctly. I assumed that I completely lost control/connection, but at this point, I can not verify if I in fact was still connected to the craft, or if the controls were reversed, or otherwise.

I also do know exactly what happened to cause the problem within those 8 seconds after takeoff, but I do know that it appears I could have detected that there was some kind of issue before I took off, which is the only point I am trying to make within this thread.

When I initially posted about this, I did not intend to draw this topic out, so I digress. I do not believe I will get concrete answers as to what exactly happened, I just wanted to share what I believe caused my crash, and how I believe this could have been detected and avoided, in hopes of saving someone else the trouble.

Good luck everyone, and I wish you nothing but safe flights ahead!

Thank you for your input. This is valuable info!

ray_gauvreau Posted at 2015-5-30 04:22
Have a look at this one and notice the 180 degree turn at 4:30 into it and the 180 degree vertical ...

Sorry for your loss... The Inspire is a great machine ...but there are still a few thing DJI need to fix.
I hope they replace it for you.

WOW!
I have had my first (and hopefully last) fly away. This may explain it. I launched and the inspire started to take off like I had full reverse stick ( Yes GPS, compass good, light winds and no high K index). I gave it an input to go forward but no response, because of my location I had already played out what I would do ...so I gave it the kill input (booth sticks down and in). I was very lucky as it did not hit anyone, amazingly the only damage was 1 broken prop (my first), the camera plate was knocked off of the vibration dampeners and 3 wires pulled out of the ribbon. I was able to fix the ribbon and test flew it the next day in a very open area with no problems. I have since flown it 4 times with no issues. When I reviewed my flight playback I can see that the inspire was not responding to my stick inputs as I would expect..

I will add orientation to my preflight checks...Thanks for you sharing!

Cheers