Terrible horizon drift - even with balanced camera and lens and starting up on flat ground.

[skylineaerial]

Hi everyone,

I'm a UK based operator who flies a number of aircraft.

I'm having real trouble with a stubborn Inspire Pro and X5 camera. I use the 12mm, 15mm and 25mm lenses.

I have a good compass calibration, good IMU calibration and always take off with a perfectly balanced camera and lens after intiliasing on a flat surface.

Despite this, I always get a horizon drift - when I fly past an object and my cameraman on the second remote pans with the object, the horizon leaves horizontal and ends up quite badly slanted. This is a very frustrating problem and I can't figure out what else to do to help with this.

Does anyone have any advice or experience?

 

[Kilrah]

Gimbal well calibrated too?

That's a problem to some extent with pretty much all 3-axis gimbals currently on the market.

 

[The Editor]

Hi everyone,

I'm a UK based operator who flies a number of aircraft.

I'm having real trouble with a stubborn Inspire Pro and X5 camera. I use the 12mm, 15mm and 25mm lenses.

I have a good compass calibration, good IMU calibration and always take off with a perfectly balanced camera and lens after intiliasing on a flat surface.

Despite this, I always get a horizon drift - when I fly past an object and my cameraman on the second remote pans with the object, the horizon leaves horizontal and ends up quite badly slanted. This is a very frustrating problem and I can't figure out what else to do to help with this.

Does anyone have any advice or experience?

I have explained this previously on the forum (if you use the search function you will find many references to the subject) but I will re-post my explanation from that thread.

"This is normal and I will try and explain why.
A three axis brushless gimbal dynamically adjusts in all three axis in real time irrespective of one particular plane of one axis - that's the whole point of it a being three axis gimbal.

How efficient the correction is will be totally dependent on the Proportional-Integral-Derivative or PID tuning of the gimbal control unit or GCU coupled with the weighting factor or 'trust' value given to the gyros over and above the accelerometers. If a high trust is put into the gyros, this will take less account of the accelerometers. This is good for slow, curved movements and will allow the gimbal to retain horizon level in those sort of maneuvers. However, if too much trust is put in the gyro then it can cause horizon drift during dynamic flight. This is why both PID and trust values are so critical. If you throw into the mix the fact that the Zenmuse X3 and X5 (like the other Zenmuse offerings) are feedback looped into the flight controller you have a very delicate balance that has to be fine tuned.
We do not know whether the algorithms within the Zenmuse are dynamically altering this 'trust' value or whether it is fixed.
In extremely high end gimbals the trust value can be set before a shot so that if the DOP is asking for a slow curving shot the parameters can be set differently to a shot that requires very fast tracking or speedy movement.

I hope that clarifies things a bit and explains why a little horizon drift is normal in use and unless DJI give us the ability to set trust weighting manually will always be there. You would have had it on your X3 but maybe were not so critical or did not notice so much. Now you are conscious of it you will be annoyed by it. "

 

[skylineaerial]

@The Editor Thanks for that, very informative.

I also use a Movi 10 for example, and it has a GPS antenna on board for aggressive moves.

I thought the Inspire would be able to do the processing itself in a way that meant it would stay level because it's hooked into the FC and the GPS/ IMU.

So really there is nothing I can do about this?

For example, if I start at the left of a tall building and fly to the right of it while my cameraman pans, I just have to sit and watch my horizon drift the whole way throughout or have my cameraman try and level it while panning smoothly?

What a disaster.

 

[RaptorMan]

I've had pretty decent results with minimal horizon tilt, but there's always some and sometime there's more.

What I do is perform the IMU calibration on a surface that is rigid and LEVEL. When performing the IMU calibration make sure that the drone has not been on long as temperature can be a problem. Also, do not walk around while doing an IMU calibration and make sure other sources of vibration including louder noise/sound is limited. Do a gimbal calibration on the same level surface.

As E. pointed out there's not a lot we can do to prevent the tilting problem but a combination of best practices for IMU and gimbal calibration along with a lighter hand on the sticks should reduce the problem to a manageable level.

This is an area that could benefit from more intensive research and development. I to would love a gimbal that was rock steady and locked parallel with gravity -- hopefully a better system will be forthcoming.


Brian