Well I've read with interest a good few fly-away / loss of control stories on here, and have to admit that with a flawless record with my own Inspire1, I've wondered if these cases could be some form of pilot error.
Well now I have experienced first hand the gut wrenching feeling of watching the craft go seemingly crazy, and feeling quite sure it would either be completely lost, or destroyed by crashing at speed.

I have a couple of hundred flights over the past 3 months, totaling probably 30 hrs, with pretty flawless performance from the Inspire1.  This event happened when I changed batteries, having flown the previous battery down to 30%, landing, replacing with a charged battery (3rd batt of the day) getting 'Green for go', and audiable 'home point updated' message, but the moment the craft lifted, it all turned to rat shit.

Although it supposedly had a full GPS lock, neutral sticks didn't give a stationary hover, it would travel at around 25mph, quickly getting to a distance where it was very hard to determine which way it was pointing. This meant I was randomely trying control stick inputs to get it under control.

The location was about as hazardous to the craft as it could get, as I was on a short beach, with 150' shear rock face behind me (30' from rock face to waters edge) so the chances of a safe recovery were slim.

The control inputs definately had an effect on the bird, so I'm sure I hadn't lost the RC Connection, the issue (I think, after going over it in my mind post-event) was that the bird was travelling at about 25mph, when it should be stationary, so control input worked, but only to try and quell the runaway speed. very hard, when you don't know which direction its actualy running away in (coz it was too distant to tell its orientation).

Incredibly, I managed to get it close to the beach, lowered the landing gear, and purely by luck, it crashed into the only soft-ish bush at the base of the rock face, doing virtually no damage. This was definately luck rather than judgement/control.

My theory is:

Upon reviewing the footage, I immediately noticed the horizon is off for this particular flight by around 15 degrees or so (never, ever seen this before with my Inspire1), I think that for whatever reason, it 'thought' it was level, when actually it was 15 degrees off, so a good part of the moment of lift from the props, is actually translated to horizontal thrust, thus, unintended horizontal speed, giving a completely unworkable scenario for the flight controller and requiring heavy control input just to maintain position (which I failed to manage for the whole 2'17" flight)

The only reason I can think of for this, is that when I changed the battery, and restarted, it was not on perfectly level ground. Could this somehow set the datum adrift for IMU??

I have heard many on here say the flyaways are due to poor compass calibration, but I can't reconcile how that could cause the behaviour I have read about, and saw on friday with my own Inspire1.

It is a truly horrible feeling watching $3k flying around over the Ocean, out of control, feeling sure it's doomed.

BTW, I tried pressing and holding the RTH button and got the beeps suggesting the command was sent, but the bird just kept traveling at high speed, so I continued fighting with control until the crash.
It all happened very quickly, and now, I'm struggling to recal exactly what happened. The time you get to try and get your head round this scenario when it occures seems like miliseconds.

Any thoughts???

I should add, that whilst losing the Inspire was a worry, It has made me very, very nervous at the thought of the harm this thing could do flying  at full speed into a crowd of people or similar!

Jon

"The only reason I can think of for this, is that when I changed the battery, and restarted, it was not on perfectly level ground. Could this somehow set the datum adrift for IMU??"

I would have to agree with you. Im glad it wasn't real bad and no got hurt.

Did you check the IMU mod values?  If you had done an IMU calibration on level surface, the Inspire should know what is level and if you take off from an uneven surface it should not matter after that calibration.  Go back through your entire set up again, reset everything to default and then reinter your settings, do an IMU calibration and then compass calibration before you fly.  If you ever see anything wrong when you initially take off, LAND.  Do not fly until you assertain what is going on.  Glad you got your Inspire back.

Tahoe_Ed Posted at 2015-7-6 05:57
Did you check the IMU mod values?  If you had done an IMU calibration on level surface, the Inspire  ...

Thanks Ed.
Didn't check mod values, as this was the third flight of the session, and it had performed perfectly prior to this batt change.

I haven't done an IMU calibration in weeks, maybe months, as it has flown flawlessly, and I don't really believe in fixing things that aren't broken (I have read on here, that religiously recalibrating, when everything is actually ok, could introduce an issue you didn't previously have, I go with that theory).  Obviously, when the last IMU calibration was done, it was on perfectly level ground.


Because I was in a fairly tight and dangerous area, I took off and accended quickly, to get safe altitude (remember, it had just flown perfectly, so I had no reason to expect this behaviour). There was no low level hover to see if all was well, so by the time I realised all was not well, I was over the salty stuff at 30+ feet, looking wide eyed and needing brown corderoy trousers!

Jon

jon@thefullers. Posted at 2015-7-6 06:09
Thanks Ed.
Didn't check mod values, as this was the third flight of the session, and it had perform ...

Jon,

I understand.  I would recommend that you do an IMU calibration on each firmware upgrade at a minimum and check your mod values on the IMU as part of your preflight check list.  I calibrate mine using this schedule and I have never had a problem.  The only time I would recommend that you do it more often is if the mod values are off or if the horizon on the gimbal is off.  It will usually cure both of those.

jon@thefullers. Posted at 2015-7-6 06:09
Thanks Ed.
Didn't check mod values, as this was the third flight of the session, and it had perform ...

A theory posed in a FB group is that turning on the Inspire while it is moving (or basically being hand held) can result in the initialization getting screwed.  I've had one occasion for which I'm certain I turned it on while carrying it to my launch point, but I don't make a habit of it certainly (and think it's likely the only time to date I've done it).  And that take off went haywire from the start.  I grounded the bird immediately and figured the area was not conducive.  But it might have been the initialization.

IMU and compass check/calibration takes only a few seconds.
Good practice to do it before flights.
I had one experience with bad IMU with the P2 and almost got in trouble in a construction site.

I'm glad everything worked out for you and you still have your I1.

My thoughts,   Check the mod value before every flight.   Shift to atti mode as soon as something isn't right.  Land it

Sorry that this happened to you.  I agree with you on the Compass Calibration.  And I'd echo what the guys above have said about the IMU values, it only takes a second to glance at them before you lift off.

The 10 second hover generally shouldn’t be a problem anywhere you can take off and land  safely. I would practice in an open area until I was comfortable hovering in confined spaces, although 30 feet by XXX of beach doesn't seem that confined.  But practice where is doesn’t matter , and don’t place yourself outside of your comfort zone, based on the A/C capability and your skills.

Also I would get familiar  with Atti mode.  I'm not saying it would or would not have helped here, but you want the option, and you need to train for it ahead of time.

Speaking of "training" we all will default to what we have trained ourselves to do, and if we haven't trained, we will probably do nothing (or at least nothing useful) when the unexpected happens.  As you pointed out a runaway Inspire is a scary thing, so a few moments of thought about what will I do today if things go haywire, are in order before each flight.  That is a good time to risk assesses the whole situation one last time too.  Maybe this day and these conditions, and this location are not the time to fly.

There is often a tilted video in conjunction with a "flyaway" but it isn't clear if the Inspire is tilted or is the camera unable to sense level, and is holding what it thinks is level with a level A/C.  Your recollection of the A/C attitude would be a better judge of the Inspire’s flight attitude.

Bottom line in my opinion is that operating an Inspire safely requires frequent proficiency training – so that is a good excuse to go out and run through some batteries – good news, right?

Hi Jon, please check the mail box, I sent you an email

Hi Jon,

There was a very high Kp-Index over the last few days. There is a lot of debate whether it has an effect or not.
Check my link: http://www.high-perspective.com/en/links/k-index.html

Thanks guys for the advice.
Ed, yes, After each update I have done a full recalibration of the IMU, so I may have exagerated re the time since last done, my point was, I subscribe to the theory held here by some, that unnecesarily disturbing a perfectly happy IMU every 5 minutes is probably a mistake. I will however, now check the figures before every flight, though it seems a worry that they're apparently so easily upset, or volatile.

I actually didn't select ATTI mode (this all happened so quick, and whilst that had always been my plan in such a scenario, in the heat of the moment I didn't) but I honestly don't think this was a gps issue, I'm pretty sure it was down to the flight controllers perception of 'level'.I have flown quite a bit in ATTI.  If my theory is correct, ATTI mode 'may' even have been harder to control it, who knows.

I'm interested though, in why the flight controller IMU settings should change within a 1 minute time period due to changing a battery.

PS. I'm also dismayed and surprised how flakey the GPS receiver seems to be (going by reports) on the Inspire, as I've been using GPS for 20 years, in powerboating (racing and pleasure) as well as Aviation (I hold a PPL) and GPS dropout is SO rare that I can't remember when I last had an issue. Jeez, A $10 bluetooth gps receiver from ebay will happily aquire and maintain a super strong signal/position inside a metal vehicle, or in a building, so why does the receiver in this very expensive device, with a superb open view of the sky have issues? I don't get it.

PPS.  Thank you Autumn, I will sort the log file later.

Kind regards

Jon

kneverett Posted at 2015-7-6 09:24
A theory posed in a FB group is that turning on the Inspire while it is moving (or basically being ...

This is a valid point ...
I sometimes carry my Inspire back to the care to change the battery ...
And I recall at one time I turned the battery on inside the care ... lifted the bird and carried it out to put it on the ground while it was "initiating" ... thinking ... wait a minute ... this is wrong ... better restart the bird while resting on the ground.

Regarding this fly away ...
As you already flew a couple of batteries ... IMU and compass calibration shouldn´t be the problem here !?
There is something else going on ... and for now I am flying my Inspire where the risk of personal injuries are minimal ... and I don´t let my kids be anywhere near this thing when the props are spinning!

The basic "quads" that you buy for 30-50 dollars ...
They don´t have compass calibration ... they don´t have advanced IMU calibration ... They don´t have "fly aways" ... just sayin!

Another thing I noticed earlier on the same day, was I sat the I1 on a surface that wasn't level, and powered it up, all seemed ok, but when I started the motors, the familiar double burst of prop rotation, followed by idle didn't happen, instead, the double burst of prop rotation was followed by a much higher idle speed, similar to what I've seen if you move, or tip the bird whilst 'idling'. it sounds like its beginning to crank up 2 motors to attempt to level the unit.
As soon as I heard this higher than usual idle, I shut it down and moved to somewhere more level (the first location was pretty crooked), and on restart, all was again normal.

When I took off from the waters side, it was as I said, not level, though nowhere near as crooked as the first location, and I didn't witness the higher motor idle speed at start up on either the good flight, or the disastrous one, so thought all was well, and indeed for the first battery duration, it was.

jon@thefullers. Posted at 2015-7-6 17:23
Another thing I noticed earlier on the same day, was I sat the I1 on a surface that wasn't level, an ...

This is good info ... thank you!

I usually leave the bird, with motors running, a few seconds on the ground (not intentionally to observe the RPM of the motors) just to focus my mind on the flight and what to do first ... take a few breaths before take off.

I will def be watching for any of this behavior now as it would be a good tell if there is something wrong with the IMU or the GPS.
Ones in the air it could obviously be to late!

Using the auto take off you never get the chance to notice this ...   
Another good reason NOT to use that!

So this is the main reason to launch from a level surface ... to avoid having the IMU trying to level the aircraft before it´s even in the air?

If the I1 is on a level surface and the IMU is trying to correct ... you know there is a problem!

Indeed, though as I said, on the last doomed flight, all seemed perfect until I was in the air thinking WTF??!!

DJI-Autumn Posted at 2015-7-6 13:59
Hi Jon, please check the mail box, I sent you an email

HI Autumn, any chance to get MOD Values adds to the home screen like while landing gear is down it shows MOD values as a pre check before flight and when the landing gear is raised it goes off, then the mod values are always present before lift off. Second would be if to make it mandatory in the shutdown process to stop all video recoding for those who forget to turn it off before powering down.

kneverett Posted at 2015-7-6 09:24
A theory posed in a FB group is that turning on the Inspire while it is moving (or basically being ...

Very interesting and it makes sense
"This event happened when I changed batteries, having flown the previous battery down to 30%, landing, replacing with a charged battery (3rd batt of the day) getting 'Green for go', and audiable 'home point updated' message, but the moment the craft lifted, it all turned to rat shit."

Jon Were you holding the Inspire in your hand (or not level) when you changed to the 3rd Batterry ?  Could explain the faulty initialization ..

arunmehta Posted at 2015-7-6 21:39
Very interesting and it makes sense
"This event happened when I changed batteries, having flown the ...

I wasn't holding it, but it was on an un-level surface at reboot.

But Tahoe Ed says the IMU calibration shouldn't be affected by this.

It's very puzzling, this I1 has never put a foot out of line before in 3 months and shed loads of flights.

I am convinced my flyaway, and many others I've read about with very similar circumstances are down to the flight controller gyro/accelerometer setup reporting a false horizon.  It makes perfect sense that this would send the thing into crazy behaviour, given that the whole principle of moving forward, backwards, sideways etc in  a multi rotor is based on tilting the bird from level, toward the direction of intended travel.

Imagine how the conflicting info to the control would mess things up, gyro reporting level flight, accelerometers and gps reporting movement.  How would it deal with that?

My crash happened on Wednesday  night,I too have had no previous trouble,as soon as it left the ground it was gone with no response from controls or fail safe,I have posted footage and flight log on the post inspire went crazy you were lucky my has been totalled hit a fence at 45mph!! I cannot get flight data as the led has been damaged,very scary to think what could have happened!! Any way check out the post

The trouble is Dji Rush out firmware updates without testing them properly I had two bricked a2s after last update 2.6.took three weeks to sort that out!!!!

jon@thefullers. Posted at 2015-7-6 23:02
I wasn't holding it, but it was on an un-level surface at reboot.

But Tahoe Ed says the IMU calibr ...

Hi Jon,

you could be correct with your theory. Here is the recovered final stages of the footage of my fly away. It clearly shows the gimbal resetting itself. This was occurring whilst craft was behaving like a wild animal. It was almost like it did the opposite of my stick inputs. With it only 3-4m in front of me, got an RC signal loss and off it went at full power backwards.

Maybe someone can explain why the gimbal resets itself like this.

Tahoe_Ed Posted at 2015-7-6 09:12
Jon,

I understand.  I would recommend that you do an IMU calibration on each firmware upgrade at a ...

Same thing happen to me! I was asked to sent my inspire 1 as DJI support weren't able to help me unless they had the inspire 1 to be able to determine the issue! that's what they said to me! and one of the landing gear broke when I was able to somehow gain some control and making a hard landing!

After getting my inspire 1 I emailed DJI support since they couldn't tell me by phone to find out what was wrong with my inspire 1. They said that they weren't able to find anything wrong and that the firmware update fixed the issue.

I am sure they saw my flight log and any information that could possibly explain what happened. Why is there no logical explanation to why so many people are experiencing the same issue?

Also I finally had the opportunity to test my inspire 1, my Gimbal was not working and a message appear saying that there was no power going to the center board that I should contact DJI for repair... I tried trouble shooting it for a while and after several attempts I decided to take out the SD card and turn my inspire 1 and it worked. I have flown my inspire 1 but for not long and I am not so confident in flying it as I am worried that it might happen during a flight! I am not so sure what will work and not work if this happens again! I am afraid that if my bird is not so close to me and I get this message and I try hitting the Return home Button that it might end up not working.  I also heard that another person in this forum had the same issue, brought down his inspire 1 and smoke came from the inspire 1. should I be alert and ground my inspire 1?? I posted a thread of my experience with that weird message.

your help will be appreciated.

Can someone tell me how the architecture is with the Inspire, i.e., the flight control sensors, such as Gyros, Accelerometers, compass etc, are they all integrated into the main board as one unit (all solid state), or, are some, or all of them at the end of cables and plugs?

If they're all integrated into one solid state unit, one would think this has the very best chance of remaining reliable with the riggors of flight, transport, crashes etc, but, if there are various sensors remotely mounted on limbs, connected via cables and plugs, this could be an issue, as cables and plugs could well introduce poor connectivity, certainly with age & use.

Any ideas guys??

Jon

Tahoe_Ed Posted at 2015-7-6 09:12
Jon,

I understand.  I would recommend that you do an IMU calibration on each firmware upgrade at a ...

I agree with you Ed. The IMU Check takes seconds to perform and is on my pre-flight checklist anytime I fly.  One occasion the numbers were off and the Inspire needed an advance IMU calibration. Not sure why it was off but I was really glad it's on my checklist.

Outlander0088 Posted at 2015-7-7 16:50
I agree with you Ed. The IMU Check takes seconds to perform and is on my pre-flight checklist anyt ...

I get that, and will obviously perform this check from now on, however, if it's as simple as a small set of numbers falling within a working tolerance before flight, the control soft/firm ware should do this check itself when you attempt to start the motors, and refuse to start them if the figures are wrong until a IMU calibration is performed? That would hardly be a tough bit of logic to build into the machine.

So many Inspire flyaway/control loss seem to be IMU related, why is this self check not part of the system, after all, the compass calibration is.
Are the IMU Mod values within spec? if yes, enable start, if no, refuse start and advise IMU calibration required.  Simples.

Tahoe_Ed Posted at 2015-7-6 09:12
Jon,

I understand.  I would recommend that you do an IMU calibration on each firmware upgrade at a ...

What value should the IMU mod be at?

jon@thefullers. Posted at 2015-7-8 02:56
I get that, and will obviously perform this check from now on, however, if it's as simple as a smal ...

You're preaching to the choir. You are correct these should be included in the standard check prior to each flight.  

How does it determine if  a compass calibration is required? Does it check to see if you are within 15 miles of your last location? The sensor values? Recent firmware update? 15 flights since the last calibration?

Like with your IMU comments, the same goes for compass. I have seen every suggestion mentioned above as a reason to perform a compass calibration. Most of the comments from DJI support.  

So why after all the comments are these not included in pre-flight checks by the system? Is someone letting the develops know these need to be included?

  

kneverett Posted at 2015-7-6 09:24
A theory posed in a FB group is that turning on the Inspire while it is moving (or basically being ...

Yeah, turning on while moving would be a bad idea I would think.

JKT Posted at 2015-7-7 08:40
Same thing happen to me! I was asked to sent my inspire 1 as DJI support weren't able to help me u ...

If I were you I would ground my Inspire and send it in to DJI.

Outlander0088 Posted at 2015-7-7 16:50
I agree with you Ed. The IMU Check takes seconds to perform and is on my pre-flight checklist anyt ...

+1. It is also on my check list.

Outlander0088 Posted at 2015-7-8 06:16
You're preaching to the choir. You are correct these should be included in the standard check prio ...

Yes we are talking with the developers in China.

tpallai Posted at 2015-7-8 03:00
What value should the IMU mod be at?

They should be 0, 1.0 and between 1400-1600.  I always do a IMU calibration after each firmware upgrade.

Here is the flight record, and video together, which I'm fairly sure I have in time together.

Interestingly, it shows a discrepancy between what the flight record shows, and what actually happened, with respect to which way the bird/camera were facing, and travelling.

There are times where the record shows the bird/cam facing the shoreline, but the video is clearly facing out to sea.  Also, there are parts where the camera clearly shows the bird tilting (and traveling as you'd expect) to the left, yet the flight record shows the opposite. As I say, I'm pretty sure the timeline of the two recordings are syncronous.
Also note the horizon well off, mostly leaning left (but not always) and most of the time, the circa 15m/s drift is to the left.

I hope some knowledge can be learned from this data.

Jon

Toilet Bowl Effect




www.crashreport.net
CrashReport - Investigation of malfunctions

Kalle Posted at 2015-7-10 18:40
Toilet Bowl Effect

How does this explain the wildly incorrect horizon shown by camera/gimble? (I have never seen this before with this inspire)  And, if compass was out for this location, why did it fly 'perfectly' for an entire battery a few minutes earlier?
I've seen the toilet bowl effect a couple of times, but only in a circular motion, not more than a 2m diameter.

The Gimbal is co-influences of the position sensors of the copter. In this way the Gimbal move is even more precise than with pure standalone gimbal. But it can also lead to incorrect information, for example by the compass.

The compass information into account in the FC in a sort of averaging routine and correct so that the long-term stability of the gyros, which have long-term drift. This correction is carried out continuously, but is particularly noticeable in a rotation of the copter noticeable because change here the compass readings.

Provides the compass comical values, which can lead to a crash or Flyaway because the FC implausible position values. In which case most conspicuously the Gimbal hangs crooked on drift direction, because it believes it would be just.

Smaller compass error does not lead to a drifting of the copter, but probably to a lopsided Gimbal. If you then are in alignment, the values ​​with the correct time and the gimbal is straight again.

www.crashreport.net
CrashReport - Investigation of malfunctions

After 56 flawless flights my started to do wild movements.  After barely getting back on the ground in one piece i sent in to DJI for checking out and repair. I was surprised when I got the following email:

Your Inspire has been inspected by the repair department. As a one time courtesy to expedite the repair process your craft has been replaced with a new unit. Any miscellaneous items like extra batteries, straps, SD cards or other personal items will be returned to you along with the new craft.
If you sent the unit without a transmitter please refer to pages 24-33 of the user manual for instructions on rebinding. Please update the firmware of the craft prior to flight. Any questions follow the link to the user manual.


Never said what was wrong, though!

Yes that's absolutely right, I've never heard the DJI had called the cause of the error and I am working for a long time with malfunctions at DJI copter.

www.crashreport.net
CrashReport - Investigation of malfunctions

I've had a chance now to go over the inspire and check a few things. Incredibly, due to crash landing in the soft bush, there was no real physical damage, but I obviously want to know 'why' it suffered this loss of control fly away.

I hear all the advice and warnings, and will of course now check IMU mod values pre flight etc, but, I checked these values today (obviously post accident) and the bird hasn't been reset, or re calibrated since the incident, and those values were spot on, so not sure why they would have gone out, caused the flyaway, then reset themselves to hide the evidence, so really not sure what to make of it.

Anyways, I recharged some batteries, recalibrated the IMU just to be sure, recalibrated the compass, and flew it.  Absolutely flawless flight, tested RTH twice, and everything performed exactly as it should (and always had previously).

Upon going over all the information I now have, (and knowing it flew perfectly a few minutes before the flyaway on a whole previous battery) I feel quite sure that had I checked the IMU pre flight, it would have shown 'good to go' and there was certainly no compass calibration warning, so I would have most likely still flown, with the same final outcome.
This leaves me very nervous about flying the bird anywhere that it could cause harm if there was a repeat episode.

The one thing that I now think is probably the best indicator that it's about to go AWOL, is the camera showing a crooked horizon. I didn't notice this pre-flight, but reviewing the video, it was as plain as the nose on your face. I've NEVER seen it have this in any pervious recordings, so the coincidence that it was present on the one flight where I experienced loss of control is just too strong to ignore, plus, I now note that many of the flyaway videos shown on this site, appear to display exactly the same issue of horribly crooked horizons. Coincidence?? I don't think so.

So in conclusion, I have no idea why the 'system' should suddenly exhibit this malfunction, lose control etc, but I am certain that the vast majority of these instances occur in conjunction with the crooked cam image, and that this means checking the camera image is level before lifting off, is probably just as, if not more important than the compass and IMU check, pre flight, as this is a very good identifier that for whatever reason, things could be about to go very bad!  I will never again lift off without first checking horizon, including panning the 360 degrees to see it level throughout, as in my video of the crash, when the gimbal does its self check thing, you can see the horizon going all over the place as it turns.  I would highly recommend everyone making this simple check, unless they want a horrible surprise to creep up on them one day.

Jon