blackcrusader Posted at 2017-10-19 19:53
You never know, my grandfather is buried in Japan, my parents were married in Japan, my cousins are half Japanese, my aunt Japanese.

Kamikaze missions

You either like flying in the clouds a lot or you just don't get much sunshine down on the ground...
Nice shots/vid
uhmmmm...maybe I should worry about you and that drone Bonsai!!!

Didn't those kamikaze guys learn their lesson once already?  *shrug*  As long as he isn't flying over Hawaii I won't worry too much, and it doesn't matter what he calls his missions.   But we do have kind of a public perception problem already,

Labroides Posted at 2017-10-19 15:33
Don't get distracted by the number of satellites - it's not having any effect on what the Phantom did.
The numbers observed are high but entirely believable.
A graph showing satellite numbers and altitude shows that numbers were high that day in that part of the world.

"Pay more attention to the thick blue line and ask yourself if that appears to be showing:"
Having control and being in control are two different things!
All the blue line shows to me is that the pilot instructed the drone to gain height and it did.

It does appear to me that the aircraft was still controllable and could have been flown back home manually, however the important question is: why did it drop into atti mode?

If it was due to the pilot operating the switch then, copying a message from one of my logs, I would expect to see:
"Flight Mode Set to Attitude__Set Flight Mode to GPS to fly safe"
instead we see:
"Satellite positioning off. Fly with Caution.",
the same message I get if there are less than 6 satellites visible and GPS is not working.

If GPS was working then why did it not do a low battery return to home?  At the end it was, according to the log, 1Km away with only 20% battery, should have returned long before.  Like blackcrusader points out, even if you tell it to fly far away in atti mode, it should still return back home unless you put some effort into defeating the RTH system and there are no messages in the log saying that RTH was overridden or was even attempted.

Looking at the log with CsvView, I see the isGPSused flag changes to false when it drops into atti, whereas in my logs using the switch to select atti the flag stays true even when in atti mode because GPS is still active and monitoring the position so that the map is accurate and it can still do an automatic low battery RTH.

CsvView also gives us a gpsLevel value which in my logs using atti mode stays at "5" at all times, but in this case it starts off at "5" but drops a few times and then drops to "1" at the same time that atti mode is entered.

Plenty to indicate that the GPS failed, only thing that suggests it didn't is the number of satellites reported is above 5, but that just indicates that it didn't fail due to the number of available satellites that it thought it had available.

Possibly from the logs it was the GPS satellite system that failed and not the drone's GPS receiver, but seems rather unlikely!

I'm not sure if the satellite count is relevant, it does seem odd that the satellite count suddenly jumps from 15 up to 21 over 2 seconds even though the aircraft is only 11m above ground and with the southern horizon hidden by a 200m high hill, but that happens 30 seconds before the change into atti mode.  In that situation I would hope to see 16 satellites, certainly wouldn't expect to see 21, and that is when flying further south where GPS satellites do actually fly overhead during some of their orbits.

Nigel_ Posted at 2017-10-20 02:26
"ay more attention to the thick blue line and ask yourself if that appears to be showing:"
Having control and being in control are two different things!
All the blue line shows to me is that the pilot instructed the drone to gain height and it did.

We may have grounds for an appeal on the verdict...

Irate Retro Posted at 2017-10-18 13:13
Since he's currently without a drone, I don't think RTH practice is what he's got on his mind at the moment.  The immediate order of business is to get a another drone.   I'd recommend the Syma X5C in this case.  Because while mode confusion might have been a contributing factor, the actual cause of the loss was the inability to fly the aircraft.

P.S. Are you sure that "pressing" RTH is enough?

I would have said the Husban H501S, solid drone that I started on.....

Genghis9 Posted at 2017-10-20 02:54
We may have grounds for an appeal on the verdict...

Grounds for an appeal, sure.  But Nigel said it himself:

"It does appear to me that the aircraft was still controllable and could have been flown back home manually"

Appeal denied.  Get a rope!

Nigel_ Posted at 2017-10-20 02:26
"Pay more attention to the thick blue line and ask yourself if that appears to be showing:"
Having control and being in control are two different things!
All the blue line shows to me is that the pilot instructed the drone to gain height and it did.

I'm tired of going over this spreadsheet.
It's a lot of work to closely study it and to write it up properly.
The OP couldn't give a toss and hasn't commented except to express displeasure at the outcome that has been confirmed by Exotherm (who I consider to be one of the best in the business)

You have ignored the important points we have both explained, that the Phantom was in full control at all times.
Every movement corresponds to joystick inputs which have been recorded.
There's nowhere in the data that can be interpreted to match the description of bringing the Phantom in to land.
And nowhere that you could say shows a Phantom going haywire and flying off into the distance.
Any ascent, forward flight or turning perfectly matches the controls performed by the pilot and none show that he made any effort to bring the Phantom back.
This does not in any way support his comment that I tried controlling the drone but the drone was not responding.
It responded perfectly to all of his control inputs.

Your comments have ignored these points which are critical to the analysis.
Instead you've concentrated on GPS count and a feeling that the GPS didn't work properly.
I'll try once more to address these- (as I have already done) but what I just mentioned above is much more important.

Any more questions, should be directed to the OP (who is used to flying until signal is lost, thinking that RTH will always save his bacon).
Ask him whether he switched to atti.
Ask him why he flew higher 6 times
Ask him to point out where he tried to bring the Phantom in to land.
Ask him why he made no effort to bring the Phantom back but appears to have concentrated on flying it higher and further away.
So far he's shown no interest in explaining or questioning anything and his initial description doesn't come close to matching what happened.

Anyway here are some comments on the points you asked about:
Having control and being in control are two different things!
All the blue line shows to me is that the pilot instructed the drone to gain height and it did.
It shows that the pilot purposely sent his Phantom up higher 6 times until it reached 750 feet.
He never made any attempt to bring it down.

the important question is: why did it drop into atti mode?
The only reason that is supported by the data is that the operator switched it to atti mode.

instead we see: "Satellite positioning off. Fly with Caution.",
the same message I get if there are less than 6 satellites visible and GPS is not working.
I don't think so.  This message usually shows when the flight controller detects something wrong with the position data.
Any position error must have only been small and momentary

Plenty to indicate that the GPS failed, only thing that suggests it didn't is the number of satellites reported is above 5, but that just indicates that it didn't fail due to the number of available satellites that it thought it had available.
You don't think that giving a valid position each 1/10th of a second is an indication that there was nothing wrong with the GPS?

I'm not sure if the satellite count is relevant, it does seem odd that the satellite count suddenly jumps from 15 up to 21 over 2 seconds even though the aircraft is only 11m above ground and with the southern horizon hidden by a 200m high hill, but that happens 30 seconds before the change into atti mode.  In that situation I would hope to see 16 satellites, certainly wouldn't expect to see 21, and that is when flying further south where GPS satellites do actually fly overhead during some of their orbits.
Look at exactly where that occurred (15:58.3).  
It's a metre or two from some tall trees to the east with others 25 metres away to the north.
Once the Phantom pops up from the cover of the trees and gets a clear view of the horizon, it can see all the satellites that were blocked.

Ignore your concerns about atti mode and GPS numbers - they are not the issue.
Look at the questions I put above - they are what this is all about.

Irate Retro Posted at 2017-10-20 03:22
Grounds for an appeal, sure.  But Nigel said it himself:

"It does appear to me that the aircraft was still controllable and could have been flown back home manually"

That was funny

Labroides Posted at 2017-10-20 03:31
I'm tired of going over this spreadsheet.
It's a lot of work to closely study it and to write it up properly.
The OP couldn't give a toss and hasn't commented except to express displeasure at the outcome that has been confirmed by Exotherm (who I consider to be one of the best in the business)

The appeal is stayed, until the defendant can provide a brief that provides answers to the applet court's questions...

In light of the stay, the punishment stands, however, no rope will be necessary...at this time

Labroides Posted at 2017-10-20 03:31
I'm tired of going over this spreadsheet.
It's a lot of work to closely study it and to write it up properly.
The OP couldn't give a toss and hasn't commented except to express displeasure at the outcome that has been confirmed by Exotherm (who I consider to be one of the best in the business)

Labroides
I know you are finding this tedious and time consuming (but you wouldn't be spending the time unless you desired to do so), it is not in vein nor a waste.
Regardless of whether this particular problem can be put to rest by all is less important as the fact that anyone following this is learning a lot from you and others.  That makes it well worth the time and effort!
Thank you!

Genghis9 Posted at 2017-10-20 03:49
Labroides
I know you are finding this tedious and time consuming (but you wouldn't be spending the time unless you desired to do so), it is not in vein nor a waste.
Regardless of whether this particular problem can be put to rest by all is less important as the fact that anyone following this is learning a lot from you and others.  That makes it well worth the time and effort!

Thanks ... I do it for the satisfaction of solving the puzzle and as an educational resource for others.
Just as in real aviation, incident investigations are very valuable educational opportunities.
It's always great to work out where a lost Phantom is and have the flyer come back the next day to say they found it close to the predicted search point or to be able to help someone with a genuine case of malfunction so they can present a case to DJI.
Occasionally you run into another type of flyer like this.

Labroides Posted at 2017-10-20 03:31
I'm tired of going over this spreadsheet.
It's a lot of work to closely study it and to write it up properly.
The OP couldn't give a toss and hasn't commented except to express displeasure at the outcome that has been confirmed by Exotherm (who I consider to be one of the best in the business)

"Any ascent, forward flight or turning perfectly matches the controls performed by the pilot"

At 16m 6.3s the yaw is at -135, the first time it goes beyond -156 is over 2 minutes later at 18m 26s,
in that time it has headed south, back north west, done 4 complete 360 degree loops, headed west, then east then north east across the road and houses while the pilot has made a few minor adjustments to the heading over a total range of +-10 degrees.  Maybe not unreasonable to describe it as haywire!

I do think it is possible that it was a magnetic problem which resulted in the GPS being disabled, but the log doesn't say anything about "magnetic" or "compass", only "Strong Interference. Fly with caution.".  If that message is related to compass errors in whatever firmware was being used then it was probably all down to taking off from near metal with a magnetic field, however it did seem to be flying OK initially and even managed to fly some ActiveTrack before running into problems.

Edit:  Actually the bit from 11m 29s when it moves away from the home point up to the point where active track is started is also haywire with the track on the map not matching the control inputs, we can assume along that bit that the track on the map is correct, "Strong Interference. Fly with caution." must mean compass problems even though it doesn't say so.  In which case the aircraft was performing correctly and it would be pilot error for ignoring the warning and ignoring the poor flight control.

Genghis9 Posted at 2017-10-20 03:46
The appeal is stayed, until the defendant can provide a brief that provides answers to the applet court's questions...

In light of the stay, the punishment stands, however, no rope will be necessary...at this time

Oops, sorry your honour,  maybe I shouldn't have spoken at that point in the proceedings?

Yep gotta agree with Genghis9 “Regardless of whether this particular problem can be put to rest by all is less important as the fact that anyone following this is learning a lot from you and others.”
Never pleasant to be droneless but explanations & processes are really worthwhile for us fly & bump people. Seems I need to dig the manual out yet again.

I agree with Genghis 9 post 50#, I have learnt a lot from this thread, sad as it is for the OP, it would be good if the AC was found at least to get more information on the flight. I am glad that there is help from forum members to try and solve these things for better or worse
Thanks guys
Eric

Nigel_ Posted at 2017-10-20 04:27
"Any ascent, forward flight or turning perfectly matches the controls performed by the pilot"

At 16m 6.3s the yaw is at -135, the first time it goes beyond -156 is over 2 minutes later at 18m 26s,

At 16m 6.3s the yaw is at -135, the first time it goes beyond -156 is over 2 minutes later at 18m 26s,
in that time it has headed south, back north west, done 4 complete 360 degree loops, headed west, then east then north east across the road and houses while the pilot has made a few minor adjustments to the heading over a total range of +-10 degrees.  Maybe not unreasonable to describe it as haywire!
Sorry ... I can't find what you are referring to here.  There's nothing in that time period to match what you describe.
All I see is the Phantom flying in a NE-E direction with some smooth curves that match the rudder inputs at:
16:33.9, 16:53.8-16:56.7, 17:37.5-17:38.2, 18:16-18:21, 18:24.7-18:29.1 and 18:32.3.
I don't see any headed south, back north west, done 4 complete 360 degree loops, headed west in there and definitely nothing like haywire.  See the screenshot
Where the Phantom is all over the place as you describe is where it is flying active track well before 16:06.3.

I do think it is possible that it was a magnetic problem which resulted in the GPS being disabled, but the log doesn't say anything about "magnetic" or "compass", only "Strong Interference. Fly with caution.".  If that message is related to compass errors in whatever firmware was being used then it was probably all down to taking off from near metal with a magnetic field, however it did seem to be flying OK initially and even managed to fly some ActiveTrack before running into problems.
The Strong Interference message refers to interference to the radio signal and has nothing to do with the compass.  If it was the compass, it would say Magnetic Field Interference.
The GPS was not disabled - it was giving location fixes every 1/10th of a second and I can't see any reason to suspect they aren't accurate.
Look at how the track perfectly matches the rudder changes mentioned above.  
There's a perfect match.
Can we put the notion that GPS was not working to bed once and for all?
There is absolutely no evidence to support it.

And for anyone saying they hope the Phantom is found ... post #12 tells us that the incident was 3 months ago so it's probably not going to be found now.

Labroides Posted at 2017-10-20 05:57
At 16m 6.3s the yaw is at -135, the first time it goes beyond -156 is over 2 minutes later at 18m 26s,
in that time it has headed south, back north west, done 4 complete 360 degree loops, headed west, then east then north east across the road and houses while the pilot has made a few minor adjustments to the heading over a total range of +-10 degrees.  Maybe not unreasonable to describe it as haywire!
Sorry ... I can't find what you are referring to here.

3 months ago ,” HE FORGOT ABOUT IT”, maybe time to forget about him. Good job anyway Labroids, good information for all in your analysis.

3 months ago, wow.  Well I do hope that the drone was found.  By a pilot, not a button-pushing photographer.

Labroides Posted at 2017-10-20 05:57
At 16m 6.3s the yaw is at -135, the first time it goes beyond -156 is over 2 minutes later at 18m 26s,
in that time it has headed south, back north west, done 4 complete 360 degree loops, headed west, then east then north east across the road and houses while the pilot has made a few minor adjustments to the heading over a total range of +-10 degrees.  Maybe not unreasonable to describe it as haywire!
Sorry ... I can't find what you are referring to here.  There's nothing in that time period to match what you describe.

Your map is a little misleading, the bottom left red dot is where it gave the "Satellite positioning off. Fly with Caution" message at 17m 34s, to get there from your 16:06.3 marker it had to go west through the loops...

"Look at how the track perfectly matches the rudder changes mentioned above. "

While it is in atti mode the track will perfectly match because the GPS positions we are given are estimated from the control inputs, they are not real GPS coordinates!

Take a break, put your spreadsheets, logfiles and dry math away for a while and try a little good old fashioned imagination and add a little human factor.
Try having a little faith in people and believe that OP tell what he has experienced even it don't match what you see from your view of the bare log.

How - simple - lets say there was two drones at play without anybody knowing.
After playing a little with the Active Track, OP look down to stop it, and do some other things,when OP look up again, he get his eyes on the second drone not realizing what is going on.
Trying to get the thing to follow controller input and it would seem out of control, going haywire, no matter how hard he try to get it one way and up i has a life of its own-
Pressing RTH a few times in panik but forget to hold it long enough while fighting, in the end he see i fly away unable to do anything.

Meanwhile he actually has sent his own drone far away and far up as in the panik he didn't notice what the camera viewed didn't match what the drone he has had eyes on would have shown.
In the end his drone is so far away that it loose connection, wait a few seconds before turning around to do the RTH but get sunblind and brought to a halt by the sensors until at some point there is only the way, down.

Even it still being a pilot error, wouldn't something like that do that both OP's experience and what the log show can be real, or did my imagination run wild?
I know by experience even a short look away can make the drone invincible even it has not moved anywhere, i know by experience that the sun can make thing do a total stop. What i don't know is if a (too) short press on RTH will show up in the log, if not i think this could be a reasonable explanation to what can have happened inclusive the Interference.

Nigel_ Posted at 2017-10-20 04:35
Oops, sorry your honour,  maybe I shouldn't have spoken at that point in the proceedings?

Blah ha ha ha

No no not at all...I've enjoyed this very interesting and highly educational exchange.  I truly have found this interesting as I feel both sides have good points, I just can't say definitively which points are the answer...but I have to tell you that I believe the bird was under full control up till it was lost.  It's the why that is problematic.  My theory is they flew it, or allowed it to fly, out so far that they could not recover it safely.  I think they flew it higher in order to ensure a direct line of sight to the antenna to boost or keep contact all the while making things worse in the process.  While I can't explain the warning messages or lack their of, I don't believe the GPS was lost, and I'm not sure what to believe referring to the switching to ATTI or not.  It seems they did switch it, but the msg question lets in doubt and again why, what were they really doing.  One thing is for sure, the story given does not match the data story being told, that is a problem.
Carry on counselor

KedDK Posted at 2017-10-20 09:49
How - simple - lets say there was two drones at play without anybody knowing.

Folks, I believe KedDK has solved it.  There was obviously a second drone on the grassy knoll.

KedDK Posted at 2017-10-20 09:49
Take a break, put your spreadsheets, logfiles and dry math away for a while and try a little good old fashioned imagination and add a little human factor.
Try having a little faith in people and believe that OP tell what he has experienced even it don't match what you see from your view of the bare log.

You would make a great mystery novel author there
However, I think enough data and info from both sides has definitively indicated that that scenario is very unlikely and most probably impossible.
We have peeled this onion so far back it is inside out now.  It has now come down to two differing opinions, while I have my suspicions and thoughts, I'm not the expert here and I can only continue to learn and read.  
I seriously doubt we will ever know for sure exactly what happened unless the operator can specifically explain what really happened and answer some key questions about the event.

Irate Retro Posted at 2017-10-20 11:09
Folks, I believe KedDK has solved it.  There was obviously a second drone on the grassy knoll.

LMAO!!!

Genghis9 Posted at 2017-10-20 11:03
Blah ha ha ha

No no not at all...I've enjoyed this very interesting and highly educational exchange.  I truly have found this interesting as I feel both sides have good points, I just can't say definitively which points are the answer...but I have to tell you that I believe the bird was under full control up till it was lost.  It's the why that is problematic.  My theory is they flew it, or allowed it to fly, out so far that they could not recover it safely.  I think they flew it higher in order to ensure a direct line of sight to the antenna to boost or keep contact all the while making things worse in the process.  While I can't explain the warning messages or lack their of, I don't believe the GPS was lost, and I'm not sure what to believe referring to the switching to ATTI or not.  It seems they did switch it, but the msg question lets in doubt and again why, what were they really doing.  One thing is for sure, the story given does not match the data story being told, that is a problem.

I think they may have taken it higher in order to keep home in view rather than getting lost over the houses, would explain the multiple increases - each time home drifted out of view, I think it was facing in approximately the right direction.

Problem now is that the response from some posters may have chased the OP away so that we wont find out what DJI conclude,  I hope the OP does post the result so that we can all learn something...

Nigel_ Posted at 2017-10-20 11:36
I think they may have taken it higher in order to keep home in view rather than getting lost over the houses, would explain the multiple increases - each time home drifted out of view, I think it was facing in approximately the right direction.

Problem now is that the response from some posters may have chased the OP away so that we wont find out what DJI conclude,  I hope the OP does post the result...

Well I agree, would very much like the OP to stay engaged so we can hear more on this from the standpoint of the companies analysis.
However, I'm not so sure it is or may be all about being chased away...IF the story was not forthright to begin with, then no amount of posting here will change that and how the OP may feel about being exposed for not being fully open about what occurred.
YET, if they are being open about everything then I would think they'd want to help try and get to the bottom of this regardless of the findings.  Conspiracy theories aside, I do think it is possible that they may have been oblivious to their situation and did things that made it worse while not realizing they were doing so.  The problem with that is that the data does not support this line of thinking.  Pilot error does not mean it was intentional but then again it could.

Yeah, sorry just couldn't stand it when the thought hit me after seeing Labroides screenshots. Very funny imagine pilot fighting to get a wrong AC to right or up.
At least one issue with the Phantoms would kill my idea, ill let it up to other to figure that out.

Nigel_ Posted at 2017-10-20 06:31
Your map is a little misleading, the bottom left red dot is where it gave the "Satellite positioning off. Fly with Caution" message at 17m 34s, to get there from your 16:06.3 marker it had to go west through the loops...

"Look at how the track perfectly matches the rudder changes mentioned above. "

Nigel ... this is getting beyond tedious.

Everything necessary to explain what happened was given in my first analysis.
I took a long time to put it together and a long time to write it up to explain clearly what happened and prevent confusion by anyone reading it.
That was then confirmed by Exotherm.

Your map is a little misleading, the bottom left red dot is where it  gave the "Satellite positioning off. Fly with Caution" message at 17m  34s, to get there from your 16:06.3 marker it had to go west through the  loops
I've politely gone back and gone over the data several times and given  even more detailed explanations for your benefit and I'm sick of it  because it's now clear that you aren't able to read the data properly.
Go and plot the positions for 16:06.3 and 18:33.9 and you'll see that they are both east of where the OP was doodling around in Active Track.

Your suggestion that the GPS positions are estimates taken from joystick positions is fanciful at best.
How does that explain the drift when the joysticks are centred?
How many times have I explained that the GPS was functioning perfectly?
Why is that so hard to accept?

All of that is just background to show the reason the Phantom did not come back.
The OP flew away downwind at high altitude where the headwind was too strong for it to be able to return.
That's all you need to know.
That's the explanation for the loss of the Phantom.
The rest is just filling in the blanks.

The nonsense you keep wanting to question does not make any difference to the outcome.
Don't waste my time any more.
Ask the AWOL OP who is responsible if you have any more questions.

Problem now is that the response from some posters may have chased the  OP away so that we wont find out what DJI conclude,  
The OP has been too embarrassed because he has been exposed for what he was doing - trying to con you and DJI
There's no escaping that conclusion when you look at the data.
There's nothing there that matches any point of his "explanation"

I hope the OP does post the result so that we can all learn something...
DJI will look at the same data we have already looked at.  
They aren't going to come to a different conclusion - they probably already did 3 months back when the OP first submitted it to them.
And they will never give an explanation as detailed as is already here.
If you want to learn something try reading what has already been posted above.

PS   You don't fly to 750 feet to keep your Phantom in sight at a distance.

Labroides Posted at 2017-10-20 13:12
Nigel ... this is getting beyond tedious.

Everything necessary to explain what happened was given in my first analysis.

"Go and plot the positions for 16:06.3 and 18:33.9 and you'll see that they are both east of where the OP was doodling around in Active Track."
Yes, that is true, and between them, the position for 16m 34.3s is most definitely west of the doodles.  

How it managed to fly those neatly curved loops between 16:06.3 and 16:34.3s while facing west all the time is unexplained but looks like a magnetic issue.

Nigel_ Posted at 2017-10-20 16:12
"Go and plot the positions for 16:06.3 and 18:33.9 and you'll see that they are both east of where the OP was doodling around in Active Track."
Yes, that is true, and between them, the position for 16m 34.3s is most definitely west of the doodles.  

I'm starting to feel like we are grasping at straws on this???
I wouldn't put any stock in how this system plots those lines on a map and invariably and quite often they are not done precisely either i.e. mis-plotted.

It may be time to agree to disagree on this...both sides have presented compelling points.  However, most have determined that the most likely outcome is that things did not play out as initially described and that throws a dark shadow over it all.  Until those questions can be resolved and a better picture of what actually occurred it will be left as unknown in the end.  However, indisputably is this, all indications say a perfectly flyable craft was flown in to the unknown either deliberately or by ignorance or unknowingly to the point it was not recoverable.  The pilot exceeded their ability to recover whether by their own fault or other but because they put the aircraft in that position it makes it pilot error, possibly with contributing factors.

Nigel_ Posted at 2017-10-20 16:12
"Go and plot the positions for 16:06.3 and 18:33.9 and you'll see that they are both east of where the OP was doodling around in Active Track."
Yes, that is true, and between them, the position for 16m 34.3s is most definitely west of the doodles.  

Sorry Nigel .. I must apologise for some of what I was thinking about you.
You are correct that 16m 34.3s is most definitely west of the doodles.
The track line gets complicated and a bit hard to follow in there.
My head hurts from getting down to this level of detail and trying to work out the 3D movements of the Phantom in a strong wind from a bunch of numbers.

How it managed to fly those neatly curved loops between 16:06.3 and 16:34.3s while facing west all the time is unexplained but looks like a magnetic issue.
Here's a graph showing the aileron and rudder inputs for that time period before the Phantom started flying in an NE-E direction (note this was almost all in P-GPS mode).
You'll see that the pilot only used a little rudder but a lot of aileron.
When using aileron, the Phantom changes direction left or right but it stays facing the same way, hence the yaw angle does not change.

Labroides Posted at 2017-10-20 18:26
Sorry Nigel .. I must apologise for some of what I was thinking about you.
You are correct that 16m 34.3s is most definitely west of the doodles.
The track line gets complicated and a bit hard to follow in there.

Yes, for an idiot like me what does this mean?

Does this mean it was haywire?
or
Does it mean the plot was off?
or
Was it...?

Genghis9 Posted at 2017-10-20 18:37
Yes, for an idiot like me what does this mean?

Does this mean it was haywire?

Quite hard to visualise what those movements of the aileron should have done, but since it was in P-GPS mode at the time so using GPS positioning, to fly a loop without using the rudder you would need a smooth sine wave on the aileron trace with half positive (above 1000) and half negative (below 1000), plus an out of phase sine wave on the elevator trace.  Those aileron movements are nothing like a smooth sine wave so could not have caused the loops on the map (going negative for 1/10th second will not produce a smooth arc).  Normally I would check what they did by looking at the x and y velocity columns, but those are calculated from the GPS so show what it actually did, they match the map, not the inputs.  My conclusion for that section of the track is that it was out of control and haywire.

Once it drops into atti mode at the end of the above graph, the stick inputs approximately match the trace plus wind drift so at that point it becomes controllable and no longer haywire but it is not using GPS positioning and the pilot appears to loose control even though it is controllable.   I don't think that the trace is plotted from the GPS at that point, the apparent sudden high speed when it temporarily re-enters P-GPS and corrects it's position to match the GPS confirms that the last part of the track is estimated.

If the log file contained some warnings of magnetic issues then it would have all made sense from the start, it doesn't but I can't see how else to explain the loops.

When it goes into atti mode, and later temporarily switches to P-GPS and back, it only gives the messages "Satellite positioning off. Fly with Caution.", there is no mention of interference of any sort which suggests that even if the loops where caused by compass issues, the loss of working GPS was nothing to do with the compass and there was some other cause of GPS failure.  

Labroides Posted at 2017-10-20 05:57
At 16m 6.3s the yaw is at -135, the first time it goes beyond -156 is over 2 minutes later at 18m 26s,
in that time it has headed south, back north west, done 4 complete 360 degree loops, headed west, then east then north east across the road and houses while the pilot has made a few minor adjustments to the heading over a total range of +-10 degrees.  Maybe not unreasonable to describe it as haywire!
Sorry ... I can't find what you are referring to here.  There's nothing in that time period to match what you describe.

Yes I did read post 12#, but I forgot just wishful thinking the drone would be found, oh well never mind, you have done well hope you keep up the good work and aren't put off
Regards
Eric

Labroides Posted at 2017-10-20 03:58
Thanks ... I do it for the satisfaction of solving the puzzle and as an educational resource for others.
Just as in real aviation, incident investigations are very valuable educational opportunities.
It's always great to work out where a lost Phantom is and have the flyer come back the next day to say they found it close to the predicted search point or to be able to help someone with a genuine case of malfunction so they can present a case to DJI.

Yes thanks for your efforts.  I remember a chap who thought his drone had been stolen in New York. He had flown it behind a building where it had landed on the rooftop. Analysis of his flight logs even showed where it should be more or less.  Low and behold he flew another drone up and there was his drone.

Not stolen and in the place where it was predicted to have landed.  

For those who refuse to accept their actions were the reason they have no drone then no amount of analysis will help.  Thick as a brick some people remain.

Nigel_ Posted at 2017-10-20 04:27
"Any ascent, forward flight or turning perfectly matches the controls performed by the pilot"

At 16m 6.3s the yaw is at -135, the first time it goes beyond -156 is over 2 minutes later at 18m 26s,

None of this is "haywire".  It is all in response to control inputs.  Each move has a corresponding manual cause.

Let's examine this segment with four 360° "loops", occurring at approximately

• 16:11.5
• 16:22
• 16:31
• 16:33
  

The image also contains a "paper plane" icon showing the orientation of the aircraft, which doesn't change during this period.  It will retain this heading unless a rudder rotation is recorded.  Sideways and forward/back movement can be performed with the right stick while maintaining this orientation.


Here is the data for the first loop.

We can see the Yaw value does not change.  We can see there is no movement of the left stick (the first two grey columns/pink headings on the right).  The yaw direction will not change unless there is rotational input -- left stick moved laterally.   The other two columns show a haphazard movement of the right stick, pulsing up and right and reflected in appropriate pitch and roll angles from the aircraft accelerometer.

Perfect data, exactly what we expect to see in a flight analysis, and the responses from the aircraft are exactly what we expect from the input instructions.  This is what a system without faults looks like.

The second "loop" also shows right stick being pulsed forward and some right input, explaining the sudden direction change, while at the same time maintaining the orientation of the aircraft in its SW heading.

The third and fourth loops, relatively close together are shown here, initiated by two right rolls.  The extent of the rolls are shown in the roll column, indicating the aircraft is tipping over around 20°-25°.  We also see for the first time any rudder input, causing the yaw direction to change from -135°  to around -143°.  This small lateral pulse from the left stick completely explains the orientation change.

NONE OF THIS INDICATES IT HAS GONE HAYWIRE!

I've never been particularly worried by the various brief warnings of compass/GPS errors.  They are trying to tell you there is a mismatch of data, that the numbers are not making sense and conflicting with data from other sensors.  They are usually brief and clear themselves once the data starts making sense again, so it is not the sign of a catastrophic failure or error.  More likely, it is probably beyond the error tolerance programmed in to the firmware and the value returned is slightly beyond what is expected.  Unless the error remains on the screen or is repeated constantly in short order, they can generally be ignored.  Many pilots here seem to give the warnings way more importance than they deserve.   If you can give us a definitive list of what each error message means, what causes it and in what circumstances, and why it was triggered, I'm all ears!  I don't think any of us can say with any certainty what causes an error to be flashed on the screen, there are way too many permutations to consider.  Just because we see "compass error", it's not indicating the compass has suddenly caught fire, it is saying the data is not making sense.  The compass heading, for example, is corroborated by data from the GPS system, and yaw/tilt/roll data from the accelerometer.  

Consider this scenario where the aircraft is travelling north.   We are all aware with the phenomenon of compass needles drifting and oscillating while moving.   The compass suddenly says "we aren't heading north any more, we are heading 10°".  The GPS sub-system says, "no, we are still incrementing at the expected positional value, the latitude/longitude values don't indicate a lateral deviation" and the accelerometer says, "we haven't deviated sideways, we are still proceeding forward normally".  Two out of three sub-systems agree, and the arbiter CPU generates a "compass error" to notify you that there wasn't a complete consensus.  It still acts on the good data from the two sub-systems and the flight continues without issue.  Milliseconds later, the compass resumes returning reliable data and the system continues operating flawlessly.  There wasn't a catastrophic failure, you haven't flown into the Twilight Zone, the government hasn't hit you with a secret wave weapon... it was just a data aberration. Don't go looking for sinister explanations to the unknown.  The simplest explanation is usually the correct explanation.

Once again, there is no evidence of GPS or other system failure, in fact the data supports that the systems were functioning correctly.  I'm not sure where you get the idea that GPS positions are calculated or estimated from inputs from the controls either.  It is just a record of the raw positional data received from the GPS device.  If it fails or is obstructed, the data will be blank or it will repeat the previous value.  If it is really screwed up it might return zeros or nonsensical values, thousands of miles away and obviously wrong, not merely inches from the last reading, 100ms ago.

In relation to the number of satellites being "too numerous", the following is the map of GPS and GLONASS satellites visible in the sky, three months ago (inferred from post #12) at midday in the vicinity of the flight.  I count 22.
Courtesy of   https://in-the-sky.org/satmap_radar.php

How could it possibly fly loop 1 without rotating the aircraft when it only ever rolls right?    The loop consists of going right, reversing a bit, going back left and then going forwards.

It would have to roll back left to bring the aircraft back again, there should be equal amounts of right roll and left roll, plus equal amounts of tilt forwards and tilt backwards to complete a smooth loop.

Nigel_ Posted at 2017-10-21 00:24
Quite hard to visualise what those movements of the aileron should have done, but since it was in P-GPS mode at the time so using GPS positioning, to fly a loop without using the rudder you would need a smooth sine wave on the aileron trace with half positive (above 1000) and half negative (below 1000), plus an out of phase sine wave on the elevator trace.  Those aileron movements are nothing like a smooth sine wave so could not have caused the loops on the map (going negative for 1/10th second will not produce a smooth arc).  Normally I would check what they did by looking at the x and y velocity columns, but those are calculated from the GPS so show what it actually did, they match the map, not the inputs.  My conclusion for that section of the track is that it was out of control and haywire.

Once it drops into atti mode at the end of the above graph, the stick inputs approximately match the trace plus wind drift so at that point it becomes controllable and no longer haywire but it is not using GPS positioning and the pilot appears to loose control even though it is controllable.   I don't think that the trace is plotted from the GPS at that point, the apparent sudden high speed when it temporarily re-enters P-GPS and corrects it's position to match the GPS confirms that the last part of the track is estimated.

So the situation is all still pure conjecture as to exactly what happened in the end?
The bird may have went haywire for a time, but then was controllable.  The GPS may have failed but could also have been turned off/turned to ATTI mode; we cannot say for certain because the messages do not seem to match the situation or messages that you'd expect to see do not appear to be there.
It remains confusing to me, and I still think the craft was controllable when lost, just not sure if the operator realized it or not.  Still think the full story has yet to be told.  
Could the spiral been the result of the bird attempting to use a smart track mode that it was not commanded to do, thus indicating a failure in the programing or circuitry?

endotherm Posted at 2017-10-22 07:28
None of this is "haywire".  It is all in response to control inputs.  Each move has a corresponding manual cause.

Let's examine this segment with four 360° "loops", occurring at approximately

Wow Labroides gives you high marks, and I can see why.
I can only make any determination about what I think happened based on the work of folks like you.  I have neither the patience or knowledge to even get close to figuring this all out.

So, to simply summarize, it is your expert opinion that the aircraft was always under full control throughout the flight, up until the time it was lost?
Understanding your views about messages, can you explain why we see neither a GPS failure warning (which there appears there were none) or that the bird was put in to ATTI mode intentionally?
Would it be safe to say, that what looks like happened is the person flew this thing in to oblivion, either knowingly or unknowingly; possibly switching to ATTI and then letting the wind assist in the loss?
Is there anything we cannot fully determine?

Thank you for your time!