Hello Everyone,

after having bothered the forum for quite a bit and quite some issues setting the copter up (took months, before it worked), many many crashes and broken equipment, i got one arducopter flying. - Unfortunately not for long as it shortly after crashed due to unknown reasons.

In the end, i decided to buy a Naza + GPS and got if flying straight away, even with different ESC, Motors, Prop's, Battery and heavily modified frame.

Naza setup and operation seems so simple to me that it's just a breeze after the issues with Arducopter.

However: I still like arducopter a lot and would prefer to use it over Naza - it offers a lot more flexibility and i just like the open-source approach.

Now, my question is:

Why is it needed to so heavily tune and configure PID's - Lot's of them - even for different flight modes etc. - while Naza simply has 6 gain settings which even work out of the box with the defaults great - with some very easy tuning (without risking or crashing the craft as on arducopter!) working even superb.

Is it maybe possible to implement this kind of control (no idea what they use, maybe self-adjusting PID?) in Arducopter ? - Or is this something which we maybe can have in PX4 ?

Best regards

Florian

Views: 770

Reply to This

Replies to This Discussion

See I did not have any of these problems.  I set the Pitch roll pid to .08 as I am using a 600mm copter and everything just works the rest are all defaults.  I get easy stabilization the waypoint navigation works fine.  Loiter is within 3 yards etc etc etc.  

I am using a motor that is a bit faster then the Arducopter ones but not by much 1350kv on a 28-26 NTM Cheap turnigy 25amp esc's and I get 6 minutes on a 2200mAh 3cell for a 1.5 KG copter. 

Hi Carl,

considering the amount of messages here on the forum with "my quad flips" etc. this shows me that there is much greater potential for trouble than with the Naza - if you search the internet for Naza flip you barely find anything.. - But - ok - this was also not my point:

My point is, is it possible to make the Arducopter settings easier or base Arducopter on the same Algo as Naza ?

This seems to be much more mature and stable - honest question: Have you tested a naza vs. arducopter in terms of stability ? - I can honestly say that i did - and there are worlds in between the performance.

So my point is: How can we make arducopter better, more stable and easier to set up ? - Is maybe using the same control technique a possibility ?

Cheers

Florian

I'm not discounting the Naza,  I was just suggesting that the users that have been reporting that their drone flips is starting off using settings that are too touchy.  I too started with the .14 PID setting and My quad flipped.  I guessed that since the PID control loop looks like a radio wave that I would approach the problem the same as radio, with the least amount of power (in this case control variable) to perform the desired function.  The worse case scenario that I have seen is that the quad is a bit sluggish to respond with a value that is too low.  With a value that is too high it starts to oscillate rapidly and then crashes.  

Hi !


Neither i am getting paid by DJI to like their stuff (in fact their stuff is expensive and closed source, as mentioned above..) -

Unfortunately, it seems that i still did not get my point through:

Naza seems to use a much more advanced regulation algo compared to the standard PID control loop stuff used here. - My point is: Could we not think about taking this approach also with Arducopter ?

It's so much better from a regulation point of view - why not look at it ?

Naza seems to use some sort of ARM CPU - so maybe this could be something more for PX4 - but should we not maybe also try to get arducopter or PX4 closer to ideal flying ?

Cheers

Florian

How would we make an algorithm similar to Naza, without having any idea what that algorithm is?  There is no magic wand for this stuff.

Hello !

I understand that it may not be that straight forward - but i assume it is still worth the effort.

All i could find so far was the following:

Advanced mix guidance algorithm and control algorithm, inherits outstanding flight stability of DJI products, provides excellent maneuverability, gives hobbyists wonderful flight experience

DJI proprietary navigational and control algorithms continue to be the best in providing excellent flight stability and highly accurate calibration of inertial locking feature to support dual-axis-gyro with PTZ control interface.


As this is a commercial product, i do not expect a design to be available somewhere public - so - i think there may be some experts here which studied control systems - i did study data technology and electronics unfortunately, so for self-adjusting PID's i am maybe not the best guy - found something here - Page 19:

http://www.honeywell.de/fp/regler/Reglerparametrierung.pdf

Even this is a temperature regulator, it is still PID - so maybe we can get some information together in the end ? - Is maybe somebody here who can guess what DJI is doing ?


 - DJI is also "only using a controller" - so i guess there is no absolute magic in there.

Cheers

Florian

Ok - followup:

I just had dinner and was thinking about the Naza-Thing again:

The naza seems to use something called "motor idle speed" to actually bring the engines close to lift off (i guess if your craft lifts off at 50%, idle speed may be 35%).

Starting from this idle speed, Naza seems to regulate. Once i had motor idle speed not set to recommended, but to low and this lead to heavy bouncing up and down. This somehow tells me that the idle speed is an offset. - I guess they use some sort of combined PID regulation algorithm with different inputs from GPS, Baro and Magnetometer internally - somehow, my guess is that they auto-calculate the PID values. - Does this help a control systems expert here further ?

Maybe introducing an offset could make the controller already much more stable due to largely reduced swings ? - Some feedback by anyone ?

Cheers

Florian

Reply to Discussion

RSS

Groups

Season Two of the Trust Time Trial (T3) Contest 
A list of all T3 contests is here. The current round, the Vertical Horizontal one, is here

© 2019   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service