Questions regarding ATTI Mode

[anom3]

I was wondering if someone could chime in on their experience in getting thrown into ATTI mode due to compass / GPS / etc errors.

I have NOT experienced this with my Mavic Pro (only about 2hr total flight time so far), but figure its a matter of time before it happens.

Worth mentioning that I have quite a bit of prior experience flying quads, the Mavic is by far not my first quad. I also avoid tight locations as I enjoy the Mavic most with my goggles flying over the large farm fields near my home. I have flown in non-GPS assisted mode on my Pixhawk based bird many times... Not fun, but I can get it back home close enough to do an emergency landing. So bare all the above in mind when answering.

So:

a) The sticks still respond (for the most part) when in ATTI mode correct? I understand the wind and such factors will basically drift the Mavic around, but doing a full throttle forward will move the bird forward, correct?

b) If unexpected and caused by compass interference / loss of GPS... Will it usually re-enter GPS mode after some time? I understand the reasons for being thrown into ATTI mode can vary... But judging by the numerous YT videos... I can expect that after some time it will hopefully regain a proper compass heading / GPS lock and let me fly the bird back for a landing?

c) When approaching the ground in ATTI mode, will OPTI mode kick in when the ground is close enough to use the visual sensors?

d) For the most part, vertical position is unaffected when in ATTI mode? My understanding is that the ATTI is essentially baro + no GPS. Some drift I suppose can be expected, but I should not expect the bird to decide to go full throttle upwards when in ATTI mode?

Final note : My "nightmare" scenario is basically loosing full control of the sticks and the bird deciding to go full throttle up. For this reason I think having the "kill motors even when flying by moving sticks to corners" turned on may not be a bad idea. While I understand its quite the expensive toy... Sure beats loosing it and having it land on someone

Thanks!

 

[Mad Mavic]

The only thing I can offer is that the Mavic, in strong enough wind, become powerless when ducking a head wind; GPS or not. I found skimming the ground effective to RTH. Carbon props do butter than stock in high winds.

It’s my onderstanding that the ultrasonic sensors are what give you altitude control an ground level, as opposed to the vision system. The optical sensors enhance horizontal positioning in well lit conditions as GPS only gets you in the ballpark.

You can practice motor on and off with the drone on the ground. Simply pull both control sticks in towards each other and both away from the antennas; towards the operator. When the drone battery is energized, the first positioning of the controls as described will simply start the motors at low rpm; much slower than in “takeoff mode” allowing the drone to sit on the ground and idle.at this point you can gently raise and lower the rpm without actually taking off if you don’t push too far forward.

Once running with props at idle speed, the same control position (in and down) will kill the motors.

I can’t imagine ever needing to do this operation in flight unless I was descending from a high altitude without enough battery for a power o controlled decent. But knowing the time required to pull the controls together, motors to energize, and full throttle up to avoid impact would be a good thing to know.

Also, unless you hear the audible “the home point has been established...” prompt at takeoff, I would advise landing.

Hope something in that helps

 

[anom3]

Yeah I have not experienced this my self but have seen this in numerous YT videos... Seams that a switch to sport mode usually does the trick.

I wonder what the max speed (angle of craft) is in ATTI mode... P-Mode or Sport mode... If its not sport mode speed then high enough winds will drift the bird just as they would in P-mode, regardless of stick input.

I guess another reason not to risk it in higher then usual winds. Which is not a problem for me, I do my best to only launch when wind conditions are low.

The only thing I can offer is that the Mavic, in strong enough wind, become powerless when ducking a head wind; GPS or not. I found skimming the ground effective to RTH. Carbon props do butter than stock in high winds.

 

[Mad Mavic]

Yeah I have not experienced this my self but have seen this in numerous YT videos... Seams that a switch to sport mode usually does the trick.

I wonder what the max speed (angle of craft) is in ATTI mode... P-Mode or Sport mode... If its not sport mode speed then high enough winds will drift the bird just as they would in P-mode, regardless of stick input.

I guess another reason not to risk it in higher then usual winds. Which is not a problem for me, I do my best to only launch when wind conditions are low.

Flying at the beach is a good example. Even in Sport mode with stock props in 15 mph winds, the MP can lose the ability to move against the wind.

I’ve crashed or lost control in almost every conceivable seniero. High winds were perhaps my most anxious. I’ve hade two flyaways with cheaper drones with auto hover, and never recovered them. The Mavic on the other hand, can power down and land regardless of your proximity to home.

Although it has cost me, I push my Mavic to the limits. My last experience with high wind was shooting video of a huge American Flag flapping in sustained 15 mph winds with gusts up to 25. I was able to do the shoot, but only after switching over to carbon props.

 

[Flawless]

ATTI mode behaviour will differ depending on the reason for it entering ATTI mode.

ATTI mode itself won't change, but how the craft behaves in ATTI mode is not just as simple as "GPS lock is off". Being prepared for ATTI mode = drift is a sure-fire way to lose your drone during a panic.

A drone that enters ATTI mode because it loses GPS lock is a simple affair. If it's near the ground, and it has downward vision enabled and functioning, it can keep position. If it's too high, it will drift with wind. Either way, it will regain GPS mode if it can.

If it is due to compass or IMU error (or for anything not GPS related), your drone will likely behave extremely erratically to any input besides throttle. Attempting to fight against wind with a drone that doesn't know which way it is heading will result in catastrophe. In situations like this, it's almost always better to land as soon as possible and in the safest possible way. Reacquiring GPS lock is extremely unlikely.

 

[anom3]

Yikes... Thanks for the input. Sounds like the IMU / Compass related ATTI mode is quite difficult to prepare for.

ATTI mode behaviour will differ depending on the reason for it entering ATTI mode.

ATTI mode itself won't change, but how the craft behaves in ATTI mode is not just as simple as "GPS lock is off". Being prepared for ATTI mode = drift is a sure-fire way to lose your drone during a panic.

A drone that enters ATTI mode because it loses GPS lock is a simple affair. If it's near the ground, and it has downward vision enabled and functioning, it can keep position. If it's too high, it will drift with wind. Either way, it will regain GPS mode if it can.

If it is due to compass or IMU error (or for anything not GPS related), your drone will likely behave extremely erratically to any input besides throttle. Attempting to fight against wind with a drone that doesn't know which way it is heading will result in catastrophe. In situations like this, it's almost always better to land as soon as possible and in the safest possible way. Reacquiring GPS lock is extremely unlikely.

 

[sar104]

ATTI mode behaviour will differ depending on the reason for it entering ATTI mode.

ATTI mode itself won't change, but how the craft behaves in ATTI mode is not just as simple as "GPS lock is off". Being prepared for ATTI mode = drift is a sure-fire way to lose your drone during a panic.

A drone that enters ATTI mode because it loses GPS lock is a simple affair. If it's near the ground, and it has downward vision enabled and functioning, it can keep position. If it's too high, it will drift with wind. Either way, it will regain GPS mode if it can.

If it is due to compass or IMU error (or for anything not GPS related), your drone will likely behave extremely erratically to any input besides throttle. Attempting to fight against wind with a drone that doesn't know which way it is heading will result in catastrophe. In situations like this, it's almost always better to land as soon as possible and in the safest possible way. Reacquiring GPS lock is extremely unlikely.

That's not correct. ATTI mode is the same in terms of flight characteristics independent of what caused it, unless the issue is actually IMU failure, which I've never seen.

In ATTI mode the FC is only stabilizing altitude and attitude. As a result it does not care about heading or course - elevator only changes pitch and aileron only changes roll, and so it will behave perfectly predictably.

In the presence of wind it will drift, just like any aircraft, and so either heading must be adjusted or aileron must be applied in order to maintain course, but that is not difficult to manage.

 

[Flawless]

Yikes... Thanks for the input. Sounds like the IMU / Compass related ATTI mode is quite difficult to prepare for.

It's not fun, for sure. If you ever lose control of your drone, your priority should be to get it on the ground before it hits someone, hits something, or flies away. That means being prepared to force it into the ground.

unless the issue is actually IMU failure, which I've never seen.

Read past the first line, Mr. Trump.

 

[sar104]

Read past the first line, Mr. Trump.

Since I was referring to your 4th paragraph, it should have been obvious that I did. Your statement:

If it is due to compass or IMU error (or for anything not GPS related), your drone will likely behave extremely erratically to any input besides throttle.​

is incorrect. Does that make it clearer?

 

[Flawless]

Since I was referring to your 4th paragraph, it should have been obvious that I did. Your statement:

If it is due to compass or IMU error (or for anything not GPS related), your drone will likely behave extremely erratically to any input besides throttle.​

is incorrect. Does that make it clearer?

Now that you've decided to quote the relevant part instead of my whole post; yes.

If the compass is in an error state that puts the drone in ATTI mode, yaw input will likely not behave as expected.

How is that incorrect?

The drone will drop GPS mode if the compass and IMU differ enough. That can simply be caused by the compass being affected by ferrous metal. You will then be in ATTI mode, but it will NOT behave as expected in this case.

It's not at all fun, but you can replicate it yourself by turning your drone on next to a large ferrous object (pipe, for example), getting GPS lock and then moving your drone away from it. Launch straight up to 5m and then try turning. Be prepared to bring it down fast.

I've done this once by accident, and once to replicate the issue. In both situations the Mavic would respond to throttle up/down reliably, but any other input would be inconsistent.

Your argument is that ATTI mode is reliable. My point is that it's the final fallback state below OPTI and GPS modes, and is demonstrably not a predictable mode of operation.

 

[sar104]

Now that you've decided to quote the relevant part instead of my whole post; yes.

If the compass is in an error state that puts the drone in ATTI mode, yaw input will likely not behave as expected.

How is that incorrect?

The drone will drop GPS mode if the compass and IMU differ enough. That can simply be caused by the compass being affected by ferrous metal. You will then be in ATTI mode, but it will NOT behave as expected in this case.

It's not at all fun, but you can replicate it yourself by turning your drone on next to a large ferrous object (pipe, for example), getting GPS lock and then moving your drone away from it. Launch straight up to 5m and then try turning. Be prepared to bring it down fast.

I've done this once by accident, and once to replicate the issue. In both situations the Mavic would respond to throttle up/down reliably, but any other input would be inconsistent.

Your argument is that ATTI mode is reliable. My point is that it's the final fallback state below OPTI and GPS modes, and is demonstrably not a predictable mode of operation.

Your description of the likely cause of switching to ATTI is correct - that is the commonest cause by far in the reports on these forums. As I explained previously, elevator and aileron will now be completely predictable because they only drive pitch and roll. In terms of rudder - the yaw motion of a quad is triggered by varying the speed of the CW vs. CCW motors and the resulting angular motion is measured by the rate gyros. The rate gyros (primary rotation data source) will still be working just fine - i.e. the FC will know the rate of rotation just fine even though it does not know the absolute heading and will be able to coordinate the rotation - and so the yaw motion will also be controlled.

What mechanism or process are you proposing is no longer working, or is different when ATTI mode is triggered by a compass, that would make the yaw uncontrolled? Note that while the FC no longer trusts the absolute heading since the sensor fusion process has failed, it still has both rate gyro and magnetometer data to detect rotation - it's not as if it can't tell how much it is turning.

In any case, disregarding the theory of how it works, you can look at log data from these events and see that the elevator, aileron and rudder inputs result in predictable maneuvering.

The uncontrolled motion that you describe is exactly what happens in that situation but it is the flight response before the FC switches to ATTI - that is the P-GPS response to an incorrect IMU yaw value.

 

[Flawless]

My primary point is that ATTI mode should not be expected to behave predictably. Certainly, in normal ATTI mode, input should have a measurable result. But ATTI mode as a fallback due to an error in flight can be caused by many things - a simple loss of GPS lock will certainly allow controlled flight. But the underlying error that causes ATTI mode to kick in may have further effects that are not immediately apparent.

I'm not denying that in ideal circumstances, it will behave as it should. But think of it like this; a drone operator prepares for an ATTI mode fallback by practising (with forced ATTI mode or another quad). Their AC suffers a fault mid-flight that puts their Mavic in ATTI mode. Connection with controller is lost, and all they have is ATTI mode on the controller display.

For the safety of their drone, themselves, and other people, they should not expect it to behave predictably. They should assume the worst - even if, at first, it does seem to respond to input as it should.

I've had it happen to me in the field, and it is not a pleasant experience. I'd rather people warn of the dangers that ATTI mode can present first, rather than say it's all fine.

 

[sar104]

My primary point is that ATTI mode should not be expected to behave predictably. Certainly, in normal ATTI mode, input should have a measurable result. But ATTI mode as a fallback due to an error in flight can be caused by many things - a simple loss of GPS lock will certainly allow controlled flight. But the underlying error that causes ATTI mode to kick in may have further effects that are not immediately apparent.

I'm not denying that in ideal circumstances, it will behave as it should. But think of it like this; a drone operator prepares for an ATTI mode fallback by practising (with forced ATTI mode or another quad). Their AC suffers a fault mid-flight that puts their Mavic in ATTI mode. Connection with controller is lost, and all they have is ATTI mode on the controller display.

For the safety of their drone, themselves, and other people, they should not expect it to behave predictably. They should assume the worst - even if, at first, it does seem to respond to input as it should.

I've had it happen to me in the field, and it is not a pleasant experience. I'd rather people warn of the dangers that ATTI mode can present first, rather than say it's all fine.

Right - but I still sort of disagree with your first statement. The point of ATTI mode is that it is entirely predictable because the pilot is in full control of pitch, roll and yaw rate, with no intervention by the FC to screw things up. The aircraft has to have failed well beyond the point of compass errors or IMU exceptions for that not to be the case.

As for your experience - I would still conclude that your aircraft had not yet switched to ATTI mode when you saw erratic behavior. But if you have a log file for that flight it would be very interesting to examine to determine what, exactly, was going on.

 

[Tinmania]

Well this thread seems to have gone off course, no pun intended.

But back to the OP’s questions I would think a simple but measured “yes to all” is correct. Sure one could say, “Not if a motor shaft is bent!” but so what, that wasn’t one of the questions.

I’m sure most here have flown a quad that doesn’t even have ATTI mode, and relied on trim to keep it somewhat in line vertically. But it still goes forward when you tell it to go forward.



Mike

 

[sar104]

Well this thread seems to have gone off course, no pun intended.

But back to the OP’s questions I would think a simple but measured “yes to all” is correct. Sure one could say, “Not if a motor shaft is bent!” but so what, that wasn’t one of the questions.

I’m sure most here have flown a quad that doesn’t even have ATTI mode, and relied on trim to keep it somewhat in line vertically. But it still goes forward when you tell it to go forward.



Mike

Yes and no - the important point is that ATTI mode is not to be feared - it is your friend if the sensor fusion scheme breaks down and the aircraft has a heading error. Without it the FC will attempt flight corrections that may drive unstable and uncontrollable flight. In ATTI the FC simply holds the aircraft level and lets the pilot do all the navigational flying. That may be tricky in windy conditions, but it is entirely predicable.

With Phantoms, ATTI practice was highly recommended, and I've no idea why DJI disabled ATTI on the Mavic. I'd mostly rather have access to ATTI than Sport, and I reprogrammed the flight mode switch accordingly.

 

[Tinmania]

Yes and no - the important point is that ATTI mode is not to be feared - it is your friend if the sensor fusion scheme breaks down and the aircraft has a heading error. Without it the FC will attempt flight corrections that may drive unstable and uncontrollable flight. In ATTI the FC simply holds the aircraft level and lets the pilot do all the navigational flying. That may be tricky in windy conditions, but it is entirely predicable.

With Phantoms, ATTI practice was highly recommended, and I've no idea why DJI disabled ATTI on the Mavic. I'd mostly rather have access to ATTI than Sport, and I reprogrammed the flight mode switch accordingly.

This is exactly what I think of ATTI mode. What I do NOT think is that it is such an emergency to warrant immediate landing no matter what (I realize you didn’t say that). I’ll fly it home just as I have other quads without gps.



Mike

 

[sar104]

This is exactly what I think of ATTI mode. What I do NOT think is that it is such an emergency to warrant immediate landing no matter what (I realize you didn’t say that). I’ll fly it home just as I have other quads without gps.

Mike

The issue with the MP and MA, at least with more recent firmware, is that the FC seems reluctant to make the switch, and a number of the reported incidents occurred while it was still flailing around in P-GPS mode trying to reconcile IMU and magnetic data. So the advice to land quickly on compass errors is conservative, but probably wise if you don't have the ability to override the FC and switch to ATTI yourself.

 

[Tinmania]

The issue with the MP and MA, at least with more recent firmware, is that the FC seems reluctant to make the switch, and a number of the reported incidents occurred while it was still flailing around in P-GPS mode trying to reconcile IMU and magnetic data. So the advice to land quickly on compass errors is conservative, but probably wise if you don't have the ability to override the FC and switch to ATTI yourself.

Again, I agree. But in that scenario we are not in ATTI mode. If the OP had asked what do when their Mavic went berserk then, yes, land asap is about it. But the questions were about already being in ATTI mode where in most cases the aircraft is perfectly controllable.

It would be nice to be able to force ATTI mode on a Mavic without modding. But my reason isn’t so I can “practice” flying in ATTI, since I don’t see that as all that hard. My reason is if I suspect GPS or compass issues I’d rather turn on ATTI mode intentionally in advance rather than let the FC make that call at a bad time.



Mike

 

[sar104]

Again, I agree. But in that scenario we are not in ATTI mode. If the OP had asked what do when their Mavic went berserk then, yes, land asap is about it. But the questions were about already being in ATTI mode where in most cases the aircraft is perfectly controllable.

It would be nice to be able to force ATTI mode on a Mavic without modding. But my reason isn’t so I can “practice” flying in ATTI, since I don’t see that as all that hard. My reason is if I suspect GPS or compass issues I’d rather turn on ATTI mode intentionally in advance rather than let the FC make that call at a bad time.

Mike

I agree that flying in ATTI is not difficult, but it is different. Drifting with the wind is one thing - one can anticipate that and conceptually understand how to compensate. Lack of positional holding and its corollary, lack of braking when sticks are centered and the need to apply reverse thrust to stop, is harder to adapt to without a little practice.

 

[Tinmania]

Indeed. But on the bright side at least you don’t have do all that and feather throttle to hold altitude.


Mike