hallmark007 Posted at 2017-10-9 12:47
I was right about nobody knows more than you, I’m glad I confused you you are just a narcissist.

The compass tells the drone which direction it's facing. The GPS data is used to find the latitude/longitude location.

Yes you absolutely confused me... funny.
How different are they? go ahead and do tell.
And after you do that, do back to my procedure and tell me where it's wrong, and considering it's iterative - tell why it can't compensate for the difference, and why letting the drone just drift blindly with the wind and hope the pilot can maneuver it in ATTI mode is a better option.
Have you even read my procedure? because if you raise the point of GPS not being able to take direction measurement while standing still, you either have not read the procedure or have not understood it.

djiuser_Kf4iPA3 Posted at 2017-10-9 12:54
Yes you absolutely confused me... funny.
How different are they? go ahead and do tell.
And after you do that, do back to my procedure and tell me where it's wrong, and considering it's iterative - tell why it can't compensate for the difference, and why letting the drone just drift blindly with the wind and hope the pilot can maneuver it in ATTI mode is a better option.

It’s simple you say it will do a RTH but how will that happen you can’t guide AC in RTH you can’t move it forward back left right, it’s automatic. And if your flying at 30 feet with obstacles in your way your system won’t work because AC will be all over the place without compass, it seems everyone is wrong no matter your right .

OK, now we're getting technical. That's better. Now we can talk actual issues and solutions.
Keep in mind that contrary to the current situation - in my procedure GPS and visual are still active, and their data is being considered. So you can use them.
Let's start with that - with GPS working (as well as baro) the AC can adjust its altitude to your regular RTH altitude. So now you run the procedure at your default RTH altitude, same as any RTH. That already reduces the chance of the AC hitting an obstacle.
Obstacle avoidance still works too, but I accept the point that it isn't perfect - even in front, thin objects would cause issues. So you can allow manual overrule, same as you allow it in regular RTH - or at least allow manual up or down to avoid obstacles. That makes sense, I accept that. It doesn't mandate throwing away the data of the systems that still work and going fully manual, becuase the iterative nature of the procedure will compensate for added errors.

BTW, why have you raised the point of GPS not knowing the direction while standing still? I know that, and my procedure doesn't use it.

djiuser_Kf4iPA3 Posted at 2017-10-9 14:04
OK, now we're getting technical. That's better. Now we can talk actual issues and solutions.
Keep in mind that contrary to the current situation - in my procedure GPS and visual are still active, and their data is being considered. So you can use them.
Let's start with that - with GPS working (as well as baro) the AC can adjust its altitude to your regular RTH altitude. So now you run the procedure at your default RTH altitude, same as any RTH. That already reduces the chance of the AC hitting an obstacle.

In order for RTH to kick in Aircraft must come to a stop then raise up then fly to homepoint then rotate to position and heading it took off from stop and then begin to landing process , so AC stops three times in RTH , you cannot control it by using sticks how can the AC know where it is when it has to continually stop..

You can pause or completely stop RTH at any time and intervene manually. The same goes here. Raise the AC further up if you have to, then resume the procedure. but in most cases the automatic rise to RTH altitude would suffice to clear obstacles on the way. All of this is the same as regular RTH.

djiuser_Kf4iPA3 Posted at 2017-10-9 14:31
You can pause or completely stop RTH at any time and intervene manually. The same goes here. Raise the AC further up if you have to, then resume the procedure. but in most cases the automatic rise to RTH altitude would suffice to clear obstacles on the way. All of this is the same as regular RTH.

So are you saying that you have to keep canceling RTH and restarting then try to get the aircraft back on track. It will eventually have to stop over homepoint but it’s likely never to get there because all the starting and stopping trying to find where it is you will have no battery left.

And one thing your leaving out is without compass any little wind will throw off its course and it’s most likely that AC when stopped without compass it will start to spin.

I think landing at a safe spot in Atti mode is a lot simpler and much less confusing for novice flyers.

I can see it now what happens when you have a compass problem.
Let go of sticks
Push stick to get heading
Hit RTH
Pause RTH
Get heading
Pause RTH
Get heading
Pause RTH
Etc etc

Hi, commercial planes are already flying, approaching and landing solely on GPS and it makes airports more efficient, ILS will eventually be totally replaced by it.
Avidyne sells certified digital auto pilots based on GPS and GPSS for  $10 to $15K + cost of installation, a bit too much for spark but hey its new technology.
All GPS units send out both analog NAV/GS signals to a CDI, and digital roll signals (GPSS) that go directly to a digital autopilot or to a roll steering converter (which converts them to analog heading commands).

Just saying the technology to fly solely on GPS already exists, any discussion about this subject will be extremely technical and be very difficult for anyone to follow.

But we are deviating from the initial intend of this thread, DJI just has to solve the compass oscillations, a compass error of 350 degrees while the difference in reality is only 10 degrees should not be that difficult to solve.

Hecate Posted at 2017-10-9 15:36
Hi, commercial planes are already flying, approaching and landing solely on GPS and it makes airports more efficient, ILS will eventually be totally replaced by it.
Avidyne sells certified digital auto pilots based on GPS and GPSS for  $10 to $15K + cost of installation, a bit too much for spark but hey its new technology.
All GPS units send out both analog NAV/GS signals to a CDI, and digital roll signals (GPSS) that go directly to a digital autopilot or to a roll steering converter (which converts them to analog heading commands).

At 10/15k it might be better to start looking at the compass.

hallmark007 Posted at 2017-10-9 15:59
At 10/15k it might be better to start looking at the compass.

Hi, yes, sure there is a reason for these compass oscillations and they do not even have to solve that, all they have to do is verify the AC position deviation from its actual position, ignore the 350 degrees error and tell its only 10 degrees kinda should do it.

Wow. This seems to show why a lot of fly away/dropped gps logs start with receiving yaw errors. Definitely seems to be pattern with this issue.

hallmark007 Posted at 2017-10-9 15:03
So are you saying that you have to keep canceling RTH and restarting then try to get the aircraft back on track. It will eventually have to stop over homepoint but it’s likely never to get there because all the starting and stopping trying to find where it is you will have no battery left.

And one thing your leaving out is without compass any little wind will throw off its course and it’s most likely that AC when stopped without compass it will start to spin.

No, that's not what I'm saying. Once again you prove you don't actually read what you're commenting about.
If the AC raises itself to regular RTH altitude, most likely you're already out of obstacles' way. If that's the case, and it would be most times, you don't need the manual intervention at all. The manual intervention is necessary only when even that's not enough, but that's the same in regular RTH. So, if the issue exists in both regular RTH and you keep advocating the necessity of RTH, then it's no reason to disqualify my procedure either. I would say that's "inconsistent" of you, to put it in understatement.

Yes, it'll be more stressful on the battery than regular RTH. It works in harder conditions, without one of its navigational systems, so it has to work harder. That's the way engineering works. But you can optimize your battery use as well. For example: the more you converge toward the right direction and your angular error decreases, the longer you can fly on each iteration - thereby, reducing the number of starts and stops and saving battery, and also collecting more GPS data in each iteration, making your collection more statistically accurate.

And no, I haven't left the wind factor out. While you were busy making technical points such as "not possible period", other people have actually raised that point already. And I've shown how, even with wind, my procedure will make the craft converge toward the right direction, including a numerical example. Once again, read before you comment.

Landing at a safe spot in ATTI is a lot simpler? Maybe for you. The pilots writing here who have lost their drones, some without even the possibility of recovering it for replacement, will beg to differ.
My procedure would have saved their drones on most occasions. And even if their drone wouldn't have made it all the way home on some occasions, it would have at least brought it closer to them - meaning less need for manual override, and higher chance of finding their drone if it doesn't make it all the way home.

Hecate Posted at 2017-10-9 16:14
Hi, yes, sure there is a reason for these compass oscillations and they do not even have to solve that, all they have to do is verify the AC position deviation from its actual position, ignore the 350 degrees error and tell its only 10 degrees kinda should do it.

Yeah... as in "the absolute value of any practical angular error along a circle cannot be over 180 deg"...
Seems like some basic applied geometry has been left out here.

djiuser_Kf4iPA3 Posted at 2017-10-9 20:54
No, that's not what I'm saying. Once again you prove you don't actually read what you're commenting about.
If the AC raises itself to regular RTH altitude, most likely you're already out of obstacles' way. If that's the case, and it would be most times, you don't need the manual intervention at all. The manual intervention is necessary only when even that's not enough, but that's the same in regular RTH. So, if the issue exists in both regular RTH and you keep advocating the necessity of RTH, then it's no reason to disqualify my procedure either. I would say that's "inconsistent" of you, to put it in understatement.

You keep saying I’m not reading maybe because this is becoming more nonsensical every time I read it. You seem to have an answer for everything , but choose to leave out what exactly happens, for instance if the Aircraft stops as will have to as per RTH it again will have no proper heading how does it get heading?
It can’t be steered by pilot as this would cancel RTH , and as I explained the likelihood  when it stops it will start to rotate at its position, as they say in my country your flogging a dead horse here, in your own admission here you say you know nothing about building Aircraft so you have decided it’s not important to take any other parts of AC into your equation and this my friend is where most of your plan falls down.
Maybe you should try to put your theory into practice and come back with the results , I think that’s where your time would be best spent. Because your starting to sound like a school teacher.

hallmark007 Posted at 2017-10-10 00:36
You keep saying I’m not reading maybe because this is becoming more nonsensical every time I read it. You seem to have an answer for everything , but choose to leave out what exactly happens, for instance if the Aircraft stops as will have to as per RTH it again will have no proper heading how does it get heading?
It can’t be steered by pilot as this would cancel RTH , and as I explained the likelihood  when it stops it will start to rotate at its position, as they say in my country your flogging a dead horse here, in your own admission here you say you know nothing about building Aircraft so you have decided it’s not important to take any other parts of AC into your equation and this my friend is where most of your plan falls down.
Maybe you should try to put your theory into practice and come back with the results , I think that’s where your time would be best spent. Because your starting to sound like a school teacher.

Yes I do feel like a schoolteacher in this argument sometimes, because your answers are those of a stubborn pupil who argues just for the sake of argument.
I say you don't read, because you keep raising points I had answered already (e.g. wind, or GPS not being able to measure direction while standing still) or saying I said things I never said (flying out of a window? really?)

And the stops during regular RTH are before and after rising up to RTH altitude, and before landing. Let's equate that with my procedure.
My procedure, as explained previously (again...) would begin after the rise to RH altitude. That's the starting point of the first iteration, so any change in yaw or position before or during that stop is taken into account already.
Then comes the iterative procedure that flies home. It's ugly, it's zigzagging, but it converges as I have already shown, including position and yaw errors introduced by wind etc. during flight or stops. They may force the procedure to perform more iterations, but it would still converge.
Furthermore - the more you get closer to your desired direction (e.g. the yaw corrections become smaller between iterations), the more you can mitigate those effects. Fly each iteration faster and over a longer distance. That will reduce the starts and stops, thereby minimize added errors and saving battery and time.
When you get home start the landing procedure. You don't have to yaw in any direction, the important thing is the position. Allow the same level of user intervention during landing as you do in regular RTH - the craft is now most likely right above the user, or very close to him/her.

And even if worse comes to worst and the craft doesn't make it all the way home, it's still better than the situation today. You have GPS readings all the way, so you immediately know your last location exactly - not the location of the compass error, but the actual last location of the craft. And since it has already flown closer to you, your radius of search is decreased. These make your chance of recovering your drone much higher than in blind drifting without GPS data.

So nothing in what you said has anything my procedure won't cover. Is it as smooth as regular RTH? No, I never said that, and it wouldn't be technically realistic to expect that when one of the navigational systems is kaput and you have to compensate for that. But if you say RTH is vital and emphesize the importance of redundancy, then how can you fight like a lion in favor of a procedure that kills redundancy and throws away valuable data? Like I said, "inconsistency", to put it mildly.

All this waffle.
Simple. DJI user is correct. If  you have GPS access, you can determine your position and after moving , direction.
Might be expensive, but theinfo is there.

interesting argument/debate here..  i think the answer lies in the middle.. dji should not at all ignore GPS data just because compass goes screwy, thats a fact, yes without compass, in short duration, gps is blind to where it is facing and going, expecting wayy to much out of such an affordable unit.. but dji could augment data better in such a event as compass failing, because as it stands right now, in a RTH event or any automated event without compass data or incorrect data, it just flys the wrong way and poof it goes, specially a spark with weaker wireless system..

Bright Spark Posted at 2017-10-10 01:31
All this waffle.
Simple. DJI user is correct. If  you have GPS access, you can determine your position and after moving , direction.
Might be expensive, but theinfo is there.

Yes your right it is waffle, simply because it’s not feasible yes you can land commercial AC with gps but at what a cost, 10/15k I’m sure if djiuser could devise his theory to work in commercial Aircraft then he could manage to save airline companies millions , maybe you know of some SUA’s that use this system or maybe you have already realized that for €500 in a tiny SUA it’s just not possible.

Agreed, expensive.
Wind would  complicate it, and obviously excessive wind will throw normal flight out of control.
Depends on processing power etc, but if it only applied short burst of -forward stick, caculate, yaw - repeatedly it could be done.

10K/15K if you add new hardware... I don't offer to add new hardware. I offer to exploit the technologies and hardware already existing on the craft in a better way, which is fully within its capabilities and potential, to achieve better results - because the outcome of today's procedure is unacceptable.

I'm with you!

HereForTheBeer Posted at 2017-10-10 02:09
interesting argument/debate here..  i think the answer lies in the middle.. dji should not at all ignore GPS data just because compass goes screwy, thats a fact, yes without compass, in short duration, gps is blind to where it is facing and going, expecting wayy to much out of such an affordable unit.. but dji could augment data better in such a event as compass failing, because as it stands right now, in a RTH event or any automated event without compass data or incorrect data, it just flys the wrong way and poof it goes, specially a spark with weaker wireless system..

The short duration is a valid technical concern.
While I agree that the shorter the duration, the bigger the error margin is - I disagree that its completely blind. There are GPS applications today that work accurately in much shorter durations (I gave an example in a previous message).
It will make the procedure work harder, sure, because the initial margin of error is bigger. But the iterative process will compensate for that over time. The bigger initial margin of error will mean harder work and more iterations, granted. So the initial duration of the first iteration should be chosen carefully - not too big, because starting out you don't know your direction yet; but big enough to achieve statistically viable GPS collection with a reasonable margin of error to start with.
And like I said, when you've corrected yourself enough - i.e. your yaw corrections become small enough, meaning your general direction is a good enough approximation - you can extend the speed and duration of each iteration, to achieve even faster and more accurate results.

If it were perfected, would the compass be obsolete?

No. And I never said it would be.
A working and accurate compass is a valuable asset, it can tell you your direction instantaneously, it can tell you your direction while you're standing still while GPS can't, it can tell you your direction while you're yawing.
I never offered to abolish it. I offer a procedure to overcome a malfunction in it, rather than let it knock down the entire navigational system.

Bright Spark Posted at 2017-10-10 03:43
If it were perfected, would the compass be obsolete?

No it won’t make compass obsolete, you will always need compass what would you do if you lose gps which is a much more common occurrence than compass problem, in fact if you look at the compass problems here on this forum most of them are almost certainly down to users,

Sorry, I thought the problem was losing decent compass causing ditching of GPS to compound your problems.Loss of GPS with no/faulty compass results in atti mode which is worst case scenario unless very close in.

No no, you were right, The issue we originated from is what happens today, that compass loss causes GPS data to be dumped even though GPS is still working.

djiuser_Kf4iPA3 Posted at 2017-10-10 05:50
No no, you were right, The issue we originated from is what happens today, that compass loss causes GPS data to be dumped even though GPS is still working.

I don’t think so , he thought the problem was loosing decent compass, not great at the reading yourself, and he is right loosing gps is much more common and just as bad a situation. Do you not think this is unacceptable ! or is it just part and parcel of flying SUA’s. IE pilot competency and knowledge will be necessary .

hallmark007 Posted at 2017-10-10 06:22
I don’t think so , he thought the problem was loosing decent compass, not great at the reading yourself, and he is right loosing gps is much more common and just as bad a situation. Do you not think this is unacceptable ! or is it just part and parcel of flying SUA’s. IE pilot competency and knowledge will be necessary .

"I thought the problem was losing decent compass causing ditching of GPS to compound your problems."
That's what he wrote. Plain English. Read first, then write.

Well costly as I see it, there's no RTH of any sort without GPS.
So ditching a serviceable GPS for errant compass is madness. At least it will hover, not drift and you may have location on the app . But it's all a bit of a compromise. Gain height may retain lost GPS, but not if compass overrules.

hallmark007 Posted at 2017-10-10 06:22
I don’t think so , he thought the problem was loosing decent compass, not great at the reading yourself, and he is right loosing gps is much more common and just as bad a situation. Do you not think this is unacceptable ! or is it just part and parcel of flying SUA’s. IE pilot competency and knowledge will be necessary .

To help each other get on the same page, I think there are now 2 different conditions being mentioned.

1. Compass error
2. GPS error

We can debate the frequency of these items, but no one here will have the definitive answer to that.

From my perspective, it seems like the bulk of this particular discussion thread has been in relation to item #1, the compass error.  I don't think anybody here has provided any discussion as to a solution for item #2.

From what I have read in this thread about the compass errors, many are making an assumption based on the logs that are seen that when a compass error occurs, DJI "CHOOSES" to turn off the GPS and ignore that data even though it appears that the GPS is still functional and could provide potentially useful data.

Can we all agree this is the fundamental starting point for this discussion?

Kloo Gee Posted at 2017-10-10 06:56
To help each other get on the same page, I think there are now 2 different conditions being mentioned.

1. Compass error

I believe so.

Bright Spark Posted at 2017-10-10 06:48
Well costly as I see it, there's no RTH of any sort without GPS.
So ditching a serviceable GPS for errant compass is madness. At least it will hover, not drift and you may have location on the app . But it's all a bit of a compromise. Gain height may retain lost GPS, but not if compass overrules.

I agree that throwing away perfectly good GPS data that you can is... well... you said it. I can think of more descriptions of the sort.
But just hovering in place, I'm afraid, is not possible. Remember, while you're standing still without a compass the craft is completely blind to where its nose is facing. So even if it realizes, through GPS, that it is drifting in a certain azimuth - it would not know where it should go to correct itself.
There is no way I can think of to avoid the need to get moving - only then will you begin to have a reference for orientation without a compass.

This is a fascinating discussion and something that I've had on my mind for quite some time when looking at the errors I see in all the logs I see on the forum.  I'm what I like to call a technical idiot!  I know just enough to think I know how everything works, but just not quite enough to know I am probably missing a key piece of knowledge that may make or break a theory.  So here goes my thoughts:

First off, DJI needs to solve the root cause of compass error whether it is a software issue or a hardware issue.  If they solve this, then most (but not all) of the reason for this discussion goes away.  

Secondly, some very important assumptions:
1. Compass, GPS, and barometer are all needed for normal operations
2. Compass, GPS, and barometer are independent systems and can provide data to system independently of each other
3. Without a working compass, the aircraft doesn't know which way it is pointed
4. When a compass error occurs, DJI "chooses" to exit out of GPS mode and go to ATTI mode reading only barometer data to hold altitude.
5. DJI could "choose" to continue utilizing GPS data when compass experiences error, but "chooses" not to in current implementation
6. Compass is required to know what direction aircraft is facing when not moving
7. A home point GPS location has been recorded and stored on the aircraft
8. As a quad-copter, the aircraft can move in any direction regardless which direction it is facing unlike a fixed wing aircraft
9. When an aircraft loses connection with the controlling device, it should initiate an RTH after documented period of time has elapsed
10. RTH already has a documented procedure where it rises to a "safe" RTH altitude and returns to home with obstacle avoidance disabled.

Can we all agree on those assumptions?
-------------------------------------------

Based on all of my assumptions being true above, here is my theory of how the problem could be solved with only GPS and barometer working, but NOT the compass.  I think a competetent programmer could make this work fairly easily.

With a known home point location marked in memory and the ability to get the current location of the aircraft with the working GPS system, the device can calculate how far away it is from the marked home point location.  However, at this point, it doesn't know what direction to initiate movement in order to get itself closer.  So at this point, it initiates a series of 4 to 8 small movements.  These movements would be something like forward, backward, left, right, forward left, forward right, backward left, and backward right.  After each of those small movements, it compares the new location with the home point and determines if it is now closer to the home point than it was previously.  

Based on these movements, the aircraft would have a rough idea of which direction to fly in order to get to the home point.  It could then calculate an imaginary line between the current location and the home point.  It then would initiate flying in the direction it roughly determined was the home point.  It then would recursively start asking itself, am I getting closer to the home point and am I staying close to the imaginary line that was calculated as the route home.  If it is getting closer to home and staying within a given deviation of the line, then continue in that direction.  If it is deviating off the line calculated for the route home, then adjust the course left or right as necessary to stay close to it.  

Now, understand that in this scenario, the aircraft may not be coming back to the home point with its face towards the home point.  But that doesn't matter at all, its not a fixed wing aircraft.   It is a quadcopter, so it can fly fairly efficiently in any direction regardless of where it is "facing".  

Also, using this method, North/South/East/West are pretty much irrelevant.  Also, using this method, it accounts for any issues with the wind trying to push it off course.

If this procedure was invoked, it should rise to the RTH altitude setting and turn off obstacle avoidance in the same way that occurs if the aircraft is more than 100 meters away when an RTH is invoked.

If the compass system returns to normal while this procedure is in progress, then it should resume normal RTH procedures instead of this special "Compass Mode Error RTH mode".

Philosophically, the question is when to initiate this Compass Error Mode RTH procedure?  Do you do it at the point today where the Spark goes into ATTI mode?  Or do you do it at the point where the remote control loses connection AND you are in ATTI mode from a Compass error?  I personally would recommend the latter, but might suggest an option in the menus to allow a user to choose.

Okay, so I put it out there, what am I missing?  

Once it calculates the imaginary line back to the home point and starts heading in that direction, it really becomes quite similar to the ol' robot following a line trick that is one of the first robotics challenges any newbie robot maker builds.  Rather than a physical line though, you are just wobbling back and forth over an imaginary line of GPS coordinates.

https://www.youtube.com/results?search_query=robot+following+line

djiuser_Kf4iPA3 Posted at 2017-10-10 06:45
"I thought the problem was losing decent compass causing ditching of GPS to compound your problems."
That's what he wrote. Plain English. Read first, then write.

Yes plain English but the problem is loosing BAD COMPASS NOT DECENT COMPASS, with decent compass you won’t lose gps unless you lose gps signal.

reality is if you fly with good compass you will not lose gps because of compass, compass problems are with regards to geo magnetic field interference flying around metal objects taking off from metal objects rebar concrete all these factors are the responsibility of the user as clearly explained in your manual.

So what you needed to do was to look at the problem that caused the bad compass and the answer to your UNACCEPTABLE PROBLEM is there, flying these dji aircraft is the responsibility of the user whether it’s €500 spark or a €35000 M210, they all operate with very similar systems, systems we are all aware of before we fly and if we are not we should be .

If the recent cause of compass problems is due to bad firmware then firmware needs to be corrected . But these problems can be wide and varied and nearly are always down to bad practices, bad practices and inexperience are pretty normal with new launches of dji drones particularly when a large amount of new users come on board. Believe me it was the very same with release of Mavic .

While you have decided to say it’s unacceptable what happens in the situation discussed, I would like to know what you think to a similar problem of aircraft dropping gps which leads to exact same problem being discussed here. “ Do you also find this unacceptable “

Kloo Gee:
I think this is the best post I've seen in a while.
First,  I think all your initial assumptions are correct. Assumption 10 is right when the craft initiates RTH more than 100m away, but when it's less than that - obstacle avoidance should work, according to the manual.

This is kinda like my procedure, except that mine focuses on flying only forward and yawing between iterations in order to keep flying forward. My rationale was to be able to use obstacle avoidance while returning.
But if obstacle avoidance is disabled, then I believe you're right - there's no point in yawing between iterations, because you don't necessarily have to face forward.
In fact, skipping the yawing would make your procedure faster and more battery-economical than mine.

Well done, sir!

Kloo Gee Posted at 2017-10-10 06:56
To help each other get on the same page, I think there are now 2 different conditions being mentioned.

1. Compass error

I fully agree, and while I’ve explained what happens in this situation, it doesn’t mean loss of aircraft and there is a redundancy for this just as there is for loss of gps in fact they are the same.
If there is a firmware causing compass problems then this should be fixed although I haven’t seen any proof that this is the case, I do believe as people gain more experience on how and where to fly these Aircraft we will see a lot less of what we are seeing for what is a very new aircraft and a lot of first time users.

i stand by i think dji is wrong in ignoring perfectly usable GPS data, theta data.  its wasteful, it should be augmented differently not ignored because the compass is freaking out..  the answer lies in the middle, i dont know what is best but the current solution i think most can agree on is both wasteful and problematic..  

like here is one simple example of how it could function.. its overly simplified but its part of data augmentation idea..  say compass is derped out, and in RTH mode and drone fly up and away, instead of continuing the wrong way it could see that its not going the right way by coordinates its crossing not being what is expected, it could then virtually attenuate the compass data rotating the drone around and then trying again coming back where it believes it should using some simple logic .. knowing which coordinates are which way then simply back tracking its incorrect move and correcting it..
or.. other solutuion is have it potion hold tight in spot with gps data, limiting or removing any drift until user slides to accept atti mode or something.. because just dropping atti mode on people and expecting them to properly react,  causes alot of issues

And I 've got to take it on the chin for being deluded enough to say it could hover sans compass.

hallmark007 Posted at 2017-10-10 07:35
Yes plain English but the problem is loosing BAD COMPASS NOT DECENT COMPASS, with decent compass you won’t lose gps unless you lose gps signal.

reality is if you fly with good compass you will not lose gps because of compass, compass problems are with regards to geo magnetic field interference flying around metal objects taking off from metal objects rebar concrete all these factors are the responsibility of the user as clearly explained in your manual.

No, the problem is losing DECENT COMPASS. Then it becomes a BAD COMPASS... and That's what he wrote
that's what starts all the trouble - losing DECENT COMPASS and HAVING BAD COMPASS.

Now that we've corrected the English, let's talk technology.
I don't aim to solve all the problems of the world. If tech A loses functionality, and it cannot be replaced by any functionality of tech B that exists on the craft, then there's no choice but to resort to manual flying. That doesn't mean you have to resort to manual flying when you DO have a potential replacement that's already supported by the existing hardware, just missing the firmware procedure. Your entire rationale of drawing symmetry between the cases is baseless, to the point of being sheer demagoguery.

I'll indulge you and share my thoughts on this nonetheless.
GPS has certain capabilities that cannot be replaced by other techs on the craft that I can think of. If you lose GPS, the IMU system (particularly the accelerometers) has the potential provide you some information for a short period of time before the error becomes too big - but if you don't reacquire GPS quickly, you'd be lost, and would have to resort to manual flying.
If you lose GPS because of objective conditions - then that's the way it is. But if you're open to the sky and still don't reacquire GPS - then there's a whole bunch of other questions that DJI has to address about the performance of Spark's GPS.

But that's beside the point, just because you asked. The point is, your approach is "if you can't fix EVERYTHING, don't fix ANYTHING". GPS has capabilities that cannot be replaced without extra hardware, which isn't practical; if the compass goes kaput, its abilities CAN be replaced. So the two issues are unrelated, nice try though...

It's acceptable to me that you resort to a bad situation when you don't have a choice; it's unacceptable that you resort to a bad situation when you DO have a choice.