Servers by jDrones

QAUTOTUNE - VTOL autotune for quadplanes


(Leonard Hall) #22

Yes, roll should behave the same as a multi-rotor and be pretty constant independent of airflow. I was specifically talking about the pitch and yaw axis (while hovering).


(Peter Hall) #23

This is my attempt to do just that for tailsitters. The thinking was that if the airspeed was high the plane mode gains work well, if the airspeed is very low then the copter mode gains work well. This just interpolates between the two based on airspeed. It seems to work well in SITL but not yet had a chance to try it in the wild.

This also allows us to use copter gains in forward flight modes for high alpha flying.

It needs a re-base but it would be great if you could give some feedback, Thanks


(Isaac ) #25

Hi!

I tried to use Autotune with my tiltrotor (4 motors, only front motors tilt). I have tested it several times with no success. The parameters for the autotune were changed (the aggressiveness from 0.075 to 0.05) but it made no difference. The first twitch (in roll) is always so hard (it reaches 50 degrees) that it destabilizes and starts losing altitude.

The weight of the plane is 9 kg and it runs T-motor 5208 340 KV with 18x8 APC propellers.

Here two logs with autotune (roll axis).

https://drive.google.com/open?id=1LicQuYhrODcgvwN1o-kFUlHV-YKBsvf-

https://drive.google.com/open?id=1jsfo99eGRz5GsbNH8IbvyDPUPp6hLfDl


(James Stewart) #26

I am having what I think is the same issue as Ismarg. Attached is a log. I tried a couple different settings tweaks in the field, such as lower aggressiveness and lower starting gains, but nothing helped. If there were a way to decrease the aggressiveness of the twitch itself, that’d likely be best.

It’s also worth nothing that I’m not sure how well the position hold aspect of it is working. Winds were 7-10mph by my guess, and the plane was getting moved around quite a bit.


(James Stewart) #27

I’ve been thinking about the issue, and the way the code is supposed to work. My best guess is that these heavy quadplanes have a high moment of inertia, which is playing havoc with the autotune. With each twitch, it’s trying to get to 20deg of rotation at 90deg/sec. The 20deg itself is probably fine, but the 90deg/sec is too fast. It causes the quadplane to overshoot, then destabilize when trying to correct.
I would change the code myself, but I am woefully inept at all things coding :open_mouth:
If anyone ekse wants to have a play with it and feels like sharing, my setup is on a Cube Black :+1:


(Hadits Baroya) #28

Hi guys,

I’m new with tailsitters plane.
i read some post in tailsitter thread, and i found someone mentioned Q_AUTOTUNE_AXES parameter.

my tailsitter already loaded with 3.6.0 arduplane (i think this is the latest firmware) , i found the QAUTOTUNE flight mode, but i cannot found the Q_AUTOTUNE_AXES . anybody knows what i’m missing to set up ?
Q_ENABLE already set to 1.


(nikhil) #29

QAutotune is not yet available in stable release. You will have to flash master in order to use qautotune. You can get latest master binary from http://firmware.ardupilot.org/Plane/latest/


(Hadits Baroya) #30

Thank you @nikhil ,
loaded the latest from the page, and i found the Q_AUTOTUNE_AXES.

where i can found the value description for this param by the way ?

the default is 7 which i assume it is for all axes.


(nikhil) #31

This param is same as copter just q is added as prefix. You can find details at http://ardupilot.org/copter/docs/autotune.html#additional-notes