A week ago I had my I1 flew away. I did use the first battery without any issue. After replacing the battery and calibrating the compass, I auto took off. The first 10 seconds the drone    flew normally. Despite that it was in the GPS mode, it suddenly started to glide to one side. I tried to return the drone to its required position to continue filming but it misbehaved. Then I initialized the return home mode. It continued flying away.

In the video I saw that my drone did not hold horizontal position as it normally does. Usually when my control sticks were in neutral positions it supposed to hold one position.  
Do you think it can be my fault for some reason?

Nomad, I'm certain there is a direct connection between the loss of level horizon seen through the camera, and the loss of control that are being reported. They seem to go hand in hand on the majority of reported flyaways.

Two weeks ago my Inspire did pretty much exactly what yours did, please see the thread here:    http://forum.dji.com/thread-23402-1-2.html

There seems to be an element of denial about the flyaway issue, but it's real, and no, having witnessed it myself, I don't think it was your fault. (before it happening to me, I was starting to suspect user error as the cause of most reported issues).

please read the thread on my flyaway and see how similar yours sounds.

Jon

jon@thefullers. Posted at 2015-7-22 01:51
Nomad, I'm certain there is a direct connection with the loss of level horizon, and the loss of cont ...

Jon, what was an official response from DJI to your complaint? Did the company replace your unit?

Nomad Posted at 2015-7-22 02:27
Jon, what was an official response from DJI to your complaint? Did the company replace your unit?

In my particular case, the inspire did crash, but incredibly,it  wasn't damaged, though obviously the question remains as to why it flew away, after countless perfect flights over the past 3 months. It could cause such serious harm to someone if it flew at them at full speed.

I have sent the mc data to DJI, as well as the Flight Record from the app. plus the videos shown on the thread, but haven't heard anything back from them as yet.

jon@thefullers. Posted at 2015-7-22 02:36
In my particular case, the inspire did crash, but incredibly,it  wasn't damaged, though obviously  ...

Do you continue using your drone? Did you fly after this?

Yes, did you not read all of the thread I linked you to?
It has subsequently flown perfectly.

My advice would be to ALWAYS check that the horizon looks level, throughout the the full 360 range of the gimbal before commiting to flight. I believe an unlevel horizon to be a strong indication things are about to go horribly wrong, DO NOT fly if you see this. restart inspire and check horizon again.

PS. why do you have at least 3 user names and locations???

I am using VPN and every time I post it shows location according to my VPN address. I did not like my previous nicknames and changed them in the menu.

When you re-calibrated the compass, you probably inadvertently replaced an originally good compass calibration with a bad calibration. After your Inspire took off into the air, away from the ground interference that caused the bad compass calibration, it is now fighting the GPS due to a bad compass calibraion, resulting in the flyaway.

I am a big advocate of finding a good remote location once and for all, get a good compass calibration and not touching it again till the next firmware update.
The compass is extremely sensitive and i never calibrate at an unknown location, and risk replacing an already good compass calibration with a bad one. From my experience, at some locations, the compass mods can go from +1800 to +1500 by just moving the Inspire a couple of feet...

How do i know i have a good calibration? After my Inspire takes off, i hover it in the air, about 3m up, make sure there's no TBE, go to the sensors page on the pilot app and review the mods. The compass mod should be right around 1500 when it is hovering in the air away from potential ground interference. Thus far, since the last firware update in May, i have yet to "re-calibrate" my compass in over a 100 flights.

frank320 Posted at 2015-7-22 03:07
When you re-calibrated the compass, you probably inadvertently replaced an originally good compass c ...

So regarding my flyaway, where it flew perfectly 5 mins earlier at the same location, and hadn't had a compass, or IMU calibration since the last update (like yourself, I buy into the 'if it aint broke, don't fix it' camp), why did mine suddenly do the same as the inspire in this thread?
If/when it happens to you, you will change your stance, just as I did...It's quite an eye opener, and certainly raises the heart rate.

jon@thefullers. Posted at 2015-7-22 03:23
So regarding my flyaway, where it flew perfectly 5 mins earlier at the same location, and hadn't ha ...

Hi Jon,
I actually followed your flyaway thread rather closely. I was thinking that your flyaway could have been related to switching on the Inspire while either moving or not on leveled ground, resulting in the IMU's mods to be non-zero on the X and Y axis during start up. If the Inspire is turned on while on leveled ground, the IMU mod values should be X=Y=0 while Z=1.0. I am guessing here, so bear with me.  

I have taken off with my Inspire on un-leveled ground before, where the IMU mods for X and Y is not 0, but some small value. What happens is that immediately after take-off, i notice that the gimbal horizon is not horizontal, tilted maybe a couple of degrees, 2-3 degrees(not 15-25 degrees like in some videos. if so, i would immediately land). But Inspire was holding position well and not TBE'ing. So i continued flying. The gimbal corrected itself to level after about 2mins of flying.

Hence, nowadays, before i do a CSC to start the rotors, i always make sure the IMU values are correct, with X=Y=0 or very close to it. I can only imagine what happens if the Inspire is tilted during take off, or fired up with the Inspire tilted...

Not trying to say that just because one checks everything before every flight means there won't be a flyaway. There's just way too many variables at play here and the only thing we can do is just eliminate as much uncertainty using available information(DJI Pilot app, mod values, level startups, etc..), and hope for the best.

Even if one did everything right, one of the many complex systems can still fail anytime.

OK, so having a saved, good IMU calibration doesn't necessarily equal level, stable flight?  It almost suggests the flight controller (at least in some part, if the camera gyros talk to the Flight controller) takes some kind of 'datum' setting at startup. It would be useful if this proves to be true, if DJI would warn of this.
I feel sure the flyaway I had was due in some way to the bird thinking it was level, when it wasn't however that may have occurred.  If indeed it was due to startup on ground that wasn't level, I guess that's good, as at least I have a reason for the loss of control, and it's something I can avoid from now on. Far better than just not knowing and just facing a mystery.

Nomad, is there any chance you started the Inspire1 for this flight, with it on un-even ground?

jon@thefullers. Posted at 2015-7-22 04:02
OK, so having a saved, good IMU calibration doesn't necessarily equal level, stable flight?  It almo ...

Jon,
I am on the same page as you, but DJI will never disclose their internal proprietary flight control algorithms, sadly. To be fair, they shouldn't, since if they do, they know it will be stolen, copied and replicated by someone else.

What i have noticed is that the flight controllers have some sort of sensor spike buffering, moving-average type control algorithm built into the flight controllers to ensure that just because one of the many sensors just received a "bad" data point, the Inspire doesn't just shoot off in a random direction just to compensate for that one bad data point. Sort of like the Kalman Filter type signal processing algorithms that are used in flight control systems.

So, what this means is that, and i am guessing here, is if for example, you started your inspire on un-leveled ground and took-off, the IMU could be internally compensating the un-leveled ground take-off to now a leveled hovering aircraft held in position by GPS, using some sort of a filter or moving-average to correct for the bias. Now, after some small time in the air, the "moving-average" corrections slide out of the calculation window and the Inspire is now incorrectly correcting a perfectly leveled aircraft using the un-leveled take-off data, and the Inspire goes haywire.

Not saying this explains all flyaways, but this could explain some cases where the flight was ok for the first 5 mins, then the inspire flies away uncontrollably, out of sight, only to return home by itself to the surprise of the owner, if the control algorithms were able to correct itself before the battery runs out.

How do it know/guess that the Inspire makes decision by filtering/moving-average? Well, when i take off from uneven ground and the gimbal horizontal is tilted on inital hover, i find that the gimbal horizontal recovers itself to perfectly level after a couple of minutes, correcting itself a little by a little over time, instead of a sudden jolt back to level horizon. In other words, the Inspire's sensors and gimbal knows that it is not leveled, and it is "slowly" correcting it.



  

That german flyaway, and subsequent return is incredible! especially how unbelievably close it comes to disaster during the flight, as recorded by the camera. Quite something, and a very lucky German chap.

Sorry to hear about your experience. Did you do a imu calibration? If so, was it on a completely level surface? If not it will show symptoms such as the inspire drifting in one direction. You should still be able to control it but it will have problems hovering steady. I've had my fair share of problems with my inspire but most were in the beginning when it was just released. Please let us know how yours is behaving now.

I had also experienced a fly-away last week. first time in 98+ flights. Luckily, after about 5 minutes of flying around and disappearing from view, it returned home and safely landed. It was pure LUCK that it didn't crash into anything, it had climbed up to 110meters clearing all nearby beach hotels.

jon@thefullers. Posted at 2015-7-22 04:37
That german flyaway, and subsequent return is incredible! especially how unbelievably close it comes ...

We have to assume there is some kind of problem here or there wouldn't be all these reports.  User error or not, the firmware is not doing a good enough job of detecting a problem situation before takeoff.  It still remains to be determined what is causing the problem though.  I wonder if DJI is actually investigating these reports seriously?  If they are, they sure are tight lipped about it.

Rob

lostinthezone Posted at 2015-7-22 08:13
We have to assume there is some kind of problem here or there wouldn't be all these reports.  Us ...

We are investigating these reports seriously. That is why we ask for logs.

BTW Jon I think your idea of a IMU self check is a good idea. I hope we implement that soon.

DJI-Dave Posted at 2015-7-22 13:51
We are investigating these reports seriously. That is why we ask for logs.

Dave, am I likely to hear back from DJI regarding my loss of control incident?  I sent everything to Autumn, as requested.
It would be good to get some feedback if nothing else, as I'm currently completely in the dark, and to be honest, a bit nervous about using the Inspire1 for fear of a repeat.

jon@thefullers. Posted at 2015-7-22 15:31
Dave, am I likely to hear back from DJI regarding my loss of control incident?  I sent everything t ...

I think you will but I really have no say or control of this.
I fully understand your being nervous due to what happened with your Inspire.