Not a member yet? Why not Sign up today
Create an account  

  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
 
PID issues with my new thrustercraft (wobbling)

#1
Hello,

Recently, I began working on a forward-broadsiding thrustercraft.

After some hard work, I managed to get it to stay relatively flat as it attacked the enemy, but the thrustercraft still had issues tilting a little to one side as it gained altitude (after spawning) and wobbling slightly as it moved.

Looking at the roll PID, I noticed that the blue line on the graph was jagged and oscillating up and down. How do I rectify this?

Screenshot of the roll PID:
[Image: fAePpv7h.png]

Attached below is the blueprint. I would appreciate some help getting this thrustercraft to stop wobbling in the air.


Attached Files
.blueprint   Project Serendipity v3.blueprint (Size: 140.99 KB / Downloads: 27)
Reply

#2
The derivative time is WAY to large. set it to something around 0.10, and work you way up to something that works. if it does not, the Gain is too large for the craft, so start it from 0.10 as well.
- - - - - -- - - - { incoming transmission...} - - - -- - - - - -

Hours: 650

Skill level: Questionable

Intelligence: Predictable

Appearance: Memorable
Reply

#3
It is syndrome of unbalanced thrust.
Try to balance it manualy(roughly)
Reply

#4
(2017-11-20, 08:32 AM)NemisisPolicy Wrote: The derivative time is WAY to large. set it to something around 0.10, and work you way up to something that works. if it does not, the Gain is too large for the craft, so start it from 0.10 as well.

Thank you for your response, I'll try what you suggested and see how that goes.

(2017-11-20, 09:59 AM)Fernir Wrote: It is syndrome of unbalanced thrust.
Try to balance it manualy(roughly)

Are you referring to the roll thrust or the yaw (sideways) thrust?
Reply

#5
Short-period overcorrection like that means the derivative controller is too active ( as was said ). If it's taking a second or two to oscillate it's probably just too much ( proportional - this function uses that for overall ) gain, and if it's several seconds it'll be the integral.
Poke my boat! mostly pre-2.0 learning & catalogue thread - Update: Heavy & light tanks 07/04/18 for 2.1. 6 ships made 2.0 aware. No more post-processing! finally! but now I can't read the forum.
Reply

#6
"Try to balance it manualy(roughly)" - it means that you first balace the body and the thrust, and it should be balanced without any PID first.
From the Depths english playlist starts here, before that it's hungarian:
https://youtu.be/Ltdx0yVI9cA?list=PLImar...ZokVtdBa73
MULTIPLAYER!

[Image: 6yFiDvF.jpg]
Reply

#7
(2017-11-23, 06:41 AM)Richard Dastardly Wrote: Short-period overcorrection like that means the derivative controller is too active ( as was said ). If it's taking a second or two to oscillate it's probably just too much ( proportional - this function uses that for overall ) gain, and if it's several seconds it'll be the integral.

Ah, okay, that makes sense now. Thank you for your explanation.

I made some changes to the roll PID in my thrustercraft since the original post and put the center of mass back to where it should be.

However, while I managed to decrease the overcorrection a little bit, the craft still wobbles slightly. I tried decreasing the integral time from "Unlimited" to some values and not much happened.

Updated BP below:


Attached Files
.blueprint   Project Serendipity v5.blueprint (Size: 171.5 KB / Downloads: 14)
Reply

#8
Integral time is the period in while the modifications are smeared in time.
If you set it to zero there will be sudden corrections and wobbling, which reduces your speed and accuracy, but enemy detection too.
From the Depths english playlist starts here, before that it's hungarian:
https://youtu.be/Ltdx0yVI9cA?list=PLImar...ZokVtdBa73
MULTIPLAYER!

[Image: 6yFiDvF.jpg]
Reply

#9
Integral value of infinite means the output is zero always, because the value is the denominator rather than the gain like every other non-self-tuning PID function. Usually ( and with every PID guide you'll find ) dt in the function is fixed and you change the gain ( usually denoted by Kp/Ki/Kd ) but not this one for some reason. Integral output of zero is what you want most of the time, it's generally not a useful controller for flight ( other than altitude ) and also generally has limits - which ours doesn't seem to have - to ensure it doesn't wind up to some astronomical value.

Fortunately Td in this function is equivalent to Kd so you can just ignore the tooltip & treat it as gain.
Poke my boat! mostly pre-2.0 learning & catalogue thread - Update: Heavy & light tanks 07/04/18 for 2.1. 6 ships made 2.0 aware. No more post-processing! finally! but now I can't read the forum.
Reply

#10
If it is at least partially self tuning PID then it is harder to detect oscillation period than proper Ki.
Reply



Forum Jump:


Users browsing this thread:
1 Guest(s)