Fly aways due to compass errors

[WithTheBirds]

I guess I am still not getting through. Let me try to explain it another way. Lets say you are driving in your car and looking at the map on the nav system. You have not programmed a route to your destination but you can see it on the map as well as a dot which indicates your position. Assume there are no roads and you don't have a compass. You just start driving in a straight line any direction. You immediately see your location dot move on the nav system and you know which way to turn to go towards your destination. As you continue to drive and watch the map on the nav system, you can quickly get pointed directly at your destination. Even if you have to drive around some obstacles or get blown a bit sideways by the wind, you can easily correct your course as long as you know your direction of travel relative to your destination. You don;t care which way is north and you don;t need to visually see your destination.
You must admit that anyone would be able to do this. A computer certainly could do it much faster and more accurately.

Have you noticed that your heading isn’t depicted on the map before you start moving? And it doesn’t always happen quickly? A simple experiment might help you appreciate what the limitations are. Open your GPS and walk or drive backwards. It will be 180 deg out. The assumption is that you are driving or walking forwards.

 

[sar104]

I guess I am still not getting through. Let me try to explain it another way. Lets say you are driving in your car and looking at the map on the nav system. You have not programmed a route to your destination but you can see it on the map as well as a dot which indicates your position. Assume there are no roads and you don't have a compass. You just start driving in a straight line any direction. You immediately see your location dot move on the nav system and you know which way to turn to go towards your destination. As you continue to drive and watch the map on the nav system, you can quickly get pointed directly at your destination. Even if you have to drive around some obstacles or get blown a bit sideways by the wind, you can easily correct your course as long as you know your direction of travel relative to your destination. You don;t care which way is north and you don;t need to visually see your destination.
You must admit that anyone would be able to do this. A computer certainly could do it much faster and more accurately.

Oh good grief - this is getting ridiculous. You are not getting through because you don't understand the basics of this subject, and what you are posting is completely incorrect. You do realize, I assume, the fundamental navigational difference between your car and drone don't you? Your car only moves in the direction that it is facing, which removes the delta between heading and track - they have to be the same. That variable is no longer variable - it's zero by definition. Aircraft, especially slow aircraft, generally have different heading and track, and no way of detecting the difference without a compass of some kind. Additionally, with a car, there is no motion except that produced by the wheels turning - there is no equivalent of drift vs vehicle motion. That is the problem that flight control has to solve.

With that I'm done trying to explain this to you, since you have clearly convinced yourself that with your completely flawed analogy you have magically discovered what has eluded everyone else, including DJI, and you are studiously ignoring every explanation posted.

 

[LivinLarge]

Have you noticed that your heading isn’t depicted on the map before you start moving? And it doesn’t always happen quickly? A simple experiment might help you appreciate what the limitations are. Open your GPS and walk or drive backwards. It will be 180 deg out. The assumption is that you are driving or walking forwards.

Exactly. When you do move, you know which way you are moving. You certainly know if you are in drive or reverse, so you know if you are going forward or backward. You also know if if you turn the wheel. In the same way, the drone knows which motors are spinning faster and which way it is tilted.

 

[Prismatic]

.

Oh good grief ...

It was once said that a billion monkeys randomly hitting the keys of a typewriter would, sooner or later, produce the complete works of Shakespeare.

The internet has proven this to be false.

 

[LivinLarge]

Oh good grief - this is getting ridiculous. You are not getting through because you don't understand the basics of this subject, and what you are posting is completely incorrect. You do realize, I assume, the fundamental navigational difference between your car and drone don't you? Your car only moves in the direction that it is facing, which removes the delta between heading and track - they have to be the same. That variable is no longer variable - it's zero by definition. Aircraft, especially slow aircraft, generally have different heading and track, and no way of detecting the difference without a compass of some kind. Additionally, with a car, there is no motion except that produced by the wheels turning - there is no equivalent of drift vs vehicle motion. That is the problem that flight control has to solve.

With that I'm done trying to explain this to you, since you have clearly convinced yourself that with your completely flawed analogy you have magically discovered what has eluded everyone else, including DJI, and you are studiously ignoring every explanation posted.

OK assume a boat instead of a car. You are on the water free to drift with the wind. You have a nav system that shows your position and the position of the destination. Are you still going to insist that you could not get there?
It is certainly possible to navigate without a compass. Skydio has proven it.

 

[sar104]

OK assume a boat instead of a car. You are on the water free to drift with the wind. You have a nav system that shows your position and the position of the destination. Are you still going to insist that you could not get there?
It is certainly possible to navigate without a compass. Skydio has proven it.

You are on the water free to drift with the wind/current? Then you are already in ATTI mode, and your boat control system has already failed. But never mind - let's take that thought further. You cannot see which way you are drifting. All you know from your moving map is that you are drifting north, for example. And let's say your intended destination is to the east. But you don't know which way is north, you don't know which way is east, and you don't know which way your boat is pointing. So which direction should you apply thrust? Forwards, backwards, left, right? You don't know is the answer. So you cannot predictively navigate, which is what flight control systems do.

What you can do is start experimenting - push forwards for example. Does your drift rate to the north increase, decrease or deviate east or west? If the resulting drift stabilizes (which it won't necessarily do if the wind/current is not constant) and you know how fast you are moving through the water/air (which you won't know accurately - you can only estimate), then you can calculate your current heading and figure out the direction to turn the boat or the thrust adjustment to produce a track to the east. But you don't have a compass, and you can only approximately figure out direction changes - the boat may be turning slowly and you can't tell, because of rate gyro drift. But let's even ignore that - perfect driftless rate gyros - and so once moving approximately east you can fine tune that solution. That's not flight control - that's reactive, trial and error navigation.

So no - I'm not going to insist that you cannot navigate by trial and error - I'm going to insist that you cannot produce a flight control system based on that. And Skydio has stated that they use optical direction finding, which serves at least one of the purposes of a compass - to detect rate gyro drift. How they figure out an absolute heading is still unexplained.

You are welcome to continue this performance without me but, since the entirety of your argument appears to be "but I can do this in my car, boat, or whatever", this is clearly about as worthwhile a use of my time as trying to explain string theory to my dogs.

 

[Prismatic]

I can imagine an optical system that generates accurate heading information under ideal conditions. But I can think of a couple of challenging situations:

1. Low-light: can I stay with my subject in the gathering twilight and then successfully fly home in near darkness?
2. Ephemeral visual fields: It's a cloudy day, and I want to film racing sailboats. My yacht--the launch point--is safely stationed a mile from a course pylon, but those are the only fixed objects within fifty miles. Is this going to work? What if shreds of fog periodically obscure one or both fixed objects?
3. Altitude effects: at high altitudes, even relatively large changes in position result in only very subtle visual changes. Can this system provide stable positioning when it's several hundred meters above the desert?

You can mix two or all three to create a really challenging scenario. I look forward to learning how well the Skydio system handles difficult situations.

 

[sar104]

I can imagine an optical system that generates accurate heading information under ideal conditions. But I can think of a couple of challenging situations:

1. Low-light: can I stay with my subject in the gathering twilight and then successfully fly home in near darkness?
2. Ephemeral visual fields: It's a cloudy day, and I want to film racing sailboats. My yacht--the launch point--is safely stationed a mile from a course pylon, but those are the only fixed objects within fifty miles. Is this going to work? What if shreds of fog periodically obscure one or both fixed objects?
3. Altitude effects: at high altitudes, even relatively large changes in position result in only very subtle visual changes. Can this system provide stable positioning when it's several hundred meters above the desert?

You can mix two or all three to create a really challenging scenario. I look forward to learning how well the Skydio system handles difficult situations.

It's easy enough to imagine how an optical system detects rotation, and the MA and M2 already have something like that. What is harder to see is how an aircraft without a real compass determines an absolute heading to begin with (i.e. initializes the IMU yaw), which is necessary for the navigation element.

 

[DanMan32]

That's the key words ...

It can from GPS delta.

 

[DanMan32]

The direction that it is facing matters because if the FC doesn't know that then it doesn't know in which direction to apply thrust. So it can plot a course and return home facing in any direction, but it has to know what that direction is.

Initially that's true, but once it starts moving, it can see which way it is moving based on GPS delta. If not the direction intended, adjust accordingly. It might not be nose first but it would be controlled.

 

[WithTheBirds]

Initially that's true, but once it starts moving, it can see which way it is moving based on GPS delta. If not the direction intended, adjust accordingly. It might not be nose first but it would be controlled.

Nobody is trying to suggest track might not be determinable by the FC employing trial and error to correlate trust inputs to change in GPS position. Read the thread- it has been agreed on numerous occasions. What is being said is that it is a stupid and impractical way of doing what becomes a trivial exercise in comparison when a compass is employed. In any case, once the FC had performed the necessary thrust/direction calculations it would know which way the nose was pointing. Who wants to have the sUAV drift or fly into the nearest obstacle while the FC performs the necessary manoeuvres to ascertain the heading? Probably not many.

For those interested in learning how Skydio flys without a compass the answers can be found in the NVIDIA white papers. The optical navigation capabilities of the current NVIDIA offering are significantly ahead of what the Movidius offerings currently provide for.

 

[sar104]

Initially that's true, but once it starts moving, it can see which way it is moving based on GPS delta. If not the direction intended, adjust accordingly. It might not be nose first but it would be controlled.

No - it cannot make a systematic adjustment because it doesn't know which way it is facing. It knows it's direction of travel, but not whether it is going forwards, backwards or sideways. It doesn't know which way to push. That only leaves the trial-and-error method, which is not a predictive control system. How many times does this have to be pointed out?

 

[DanMan32]

As a fail-safe, it doesn't need to know which way the nose is facing, only how to get to the target GPS coordinates. Even then, based on thrust applied to accomplish the track, it can figure out the direction it's pointing. After all, even in ATTI mode, FC knows how to apply thrust when you tell it to go forward.

 

[DanMan32]

Nobody is trying to suggest track might not be determinable by the FC employing trial and error to correlate trust inputs to change in GPS position. Read the thread- it has been agreed on numerous occasions. What is being said is that it is a stupid and impractical way of doing what becomes a trivial exercise in comparison when a compass is employed. In any case, once the FC had performed the necessary thrust/direction calculations it would know which way the nose was pointing. Who wants to have the sUAV drift or fly into the nearest obstacle while the FC performs the necessary manoeuvres to ascertain the heading? Probably not many.

For those interested in learning how Skydio flys without a compass the answers can be found in the NVIDIA white papers. The optical navigation capabilities of the current NVIDIA offering are significantly ahead of what the Movidius offerings currently provide for.

I think you are mistakening regular flight algorithm vs recovery on error. Rather than go berserk when a compass discrepancy is detected, use GPS to correct for it or at least fail-safe landing or RTH.
Of course, use current means using compass when everything agrees.

 

[sar104]

As a fail-safe, it doesn't need to know which way the nose is facing, only how to get to the target GPS coordinates. Even then, based on thrust applied to accomplish the track, it can figure out the direction it's pointing. After all, even in ATTI mode, FC knows how to apply thrust when you tell it to go forward.

I'm sorry - but this discussion has become pointless. You are not getting this at all.

 

[WithTheBirds]

I think you are mistakening regular flight algorithm vs recovery on error. Rather than go berserk when a compass discrepancy is detected, use GPS to correct for it or at least fail-safe landing or RTH.
Of course, use current means using compass when everything agrees.

You seem to be assuming there is some magical difference in the FC controlling propulsion outputs to fly a particular path between user stick inputs or a failsafe action. For the purpose of this discussion that is probably an irrelevant discussion.

Rather than continuing with what seems to be a series of attempts to rehabilitate a bad argument try going back to the core problem- how does the AC determine heading without a compass? Trial and error propulsion input/change in GPS position is not a practical consideration.

 

[WithTheBirds]

As a fail-safe, it doesn't need to know which way the nose is facing, only how to get to the target GPS coordinates. Even then, based on thrust applied to accomplish the track, it can figure out the direction it's pointing. After all, even in ATTI mode, FC knows how to apply thrust when you tell it to go forward.

I would be very surprised to learn the compass isn't used in ATTI mode- from my limited understanding ATTI simply ignores for the purpose of position hold.

 

[DanMan32]

I would be very surprised to learn the compass isn't used in ATTI mode- from my limited understanding ATTI simply ignores for the purpose of position hold.

Compass error is one reason a DJI system would go into ATTI mode. The only stabilization you get in that mode is height based on barometer.
Now if ATTI is engaged voluntarily, then GPS and compass would be used for telemetry to the user, but not for flight control.

 

[sar104]

Compass error is one reason a DJI system would go into ATTI mode. The only stabilization you get in that mode is height based on barometer.
Now if ATTI is engaged voluntarily, then GPS and compass would be used for telemetry to the user, but not for flight control.

ATTI stabilizes pitch, roll, yaw and altitude. Just not position.

 

[Prismatic]

Back to the boat. You have GPS but no compass. Dense fog. Strong currents that vary dramatically with the tide, and you without a tide table. Shoals every direction but due south. Your heading is unknown, but GPS shows you’re dangerously close to a reef, and moving that way. It could be ahead, astern, port, starboard, no telling.

What’s the plan?