Usual list of timestamps: 0:29 AI PID vs GP PID 4:25 PID variables 6:15 Gain 8:05 Derivative 9:56 Integral 12:45 Why use integral 13:21 Tuning 16:15 Troubleshooting
This video is amazing, thank you, ive been using PID for ages but when i turn off AI they stop working, i didnt realise there were GP PID's this is exactly what i was looking for, thank you
Things to note what i found up by tests and how i understand it: GAIN - % of propulsion linked with PID it has AUTHORITY over, the higher the less force propulsion has in certain situations when PID try to smooth agile maneuever out. DERIVATIVE - Good for tweaking, this "prediction" acts like buffer and accelerator - it will help with recoil if your ship wobble after fire and can help with to rapid as well as slugish movement - WARNING, high values can INCREASE power consumption, while giving no adventage, test not only in calm but movement scenarios and set it so ship regains balance fast. INTEGRAL - Is a limiter that forces PID to apply it's fix in set period of time. High value gives more time for fix and is more fluid (large bombardment platforms) while low values force rapid fix (small agile gunships). BASED UPON propulsion placement, ammounth and is it roll/pitch/yaw/hover/strafe PID values will differ horribly. Also using propulsion for both ...examp. Hover and Pitch...is questionable sometimes. Atmosphere and Space settings are hard to combine, also AI maneuever/Behaviour and center of mass will have their say, so every change on ship needs PID check. ONCE AGAIN ...stabilizing ship in calm situation is not enough, if force/impulse is applied it need to regain balance, also unproper setting might give you "stabilization" in both situations while power consumption might be higer by 200% of what's actually needed !!! (What can looks like "stable" and "calm" visualy in game FX in reality in game code those engines can fire for example left/right constantly trying to even themself out).
A lot of people have their own theories about PIDs and are unfortunately wrong. I can't fault people, it's not information that is readily available and there's details I don't quite get even now. That said, a PID without a gain of 1 can use 100% of the available thrust depending on the difference between the setpoint and the current state of the vehicle. Integral also "accumulates" error over time rather than forcing the PID to fix the error within that time though I can see why you'd think that way given a PID with low integral will push quite hard quite quickly if it's struggling to stabilise the vehicle. As for not allowing each thruster to control every possible axis it can control, it's a whole other debate and there's pros and cons to each approach. However, I am quite confident when I say that enabling more than one axis for most thrusters is the more efficient option in the end, but it does require careful tuning and to plan ahead for situations you might not even imagine before you see it happen.
If you're familiar with a spring-mass-damper system, essentially the proportional component is like a spring and the derivative component is like a damper. Here, the proportional gain is equivalent to the spring constant. In the form that's in the game, the damping coefficient is equivalent to the product of the proportional gain and the derivative time. I don't want to get into the math of the integral component, but just know that in the form present in the game its relative influence is proportional to the ratio of proportional gain to integral time. I would avoid using integral control unless you have steady-state error you need to get rid of. Besides eliminating steady-state error, when your set point changes integral control will decrease the time it takes to get there but will then cause overshoot due to the accumulation of error.
I'm aware that some of the specifics of how I explained things do not necessarily fully line up with the math. That said, I think it does a decent job of explaining it as simply as possible for the less math-inclined people. As far as integral goes, yes, it can cause overshoot. However, for combat vehicles where thrust is likely to be damaged, integral is a way to maintain control for longer. For that reason, I actually encourage people to use it even if they don't need to.
You summed up what I would have commented otherwise! Accumulation of error on the I term is an issue in this game, which makes PD controllers (I set to 250s) a safer and just slightly less accurate fallback. I only ever use the I term on altitude control, and pitch for frontsiders that may need it.
No joke i came here to understand control diagrams for my engineering course better lol, ftd players know their way to make complex concepts like this seem easier
Your tutorials are really top notch, Ohm! No doubts they will be invaluable to newcomers, I know they would have saved me a lot of hassle back when I started playing this game. Do you plan to make a part 3 to the Breadboard series?
I've been looking for a tutorial that explained the equations variables instead of telling what they use on their particular craft. My thanks for the clarifications. I can do a lot more with an understood problem vs than a given answer.
Oh my god I finally found a video that helped me understand how to tweak a PID! Thank you so much, the difference between my pid craft and the ones I see on UA-cam was driving me insane!
What the times "actually" mean, in as simple terms as I can make it, is: Integral looks at the average of the error over the last integral time seconds, and tries to correct that. Derivative predicts how big the error will be in derivative time seconds, and tries to correct that.
Thank you for your feelings. I don't think it's fair to say I should have the monopoly on tutorials. I am a trained teacher which helps, but FtD is complex and different people will prefer different approaches to explaining things. I'm not proud that I am so harsh sometimes when people mean well, but I am indeed passionate about educational material... sometimes it's good, sometimes I should take a chill pill. ¯\_(ツ)_/¯
It should work fine for ships, though the ratio between thrust-to-weight and water resistance will make tweaking things harder. Also, yaw tends to be more finicky just in general and I would definitely recommend against using integral for yaw.
Usual list of timestamps:
0:29 AI PID vs GP PID
4:25 PID variables
6:15 Gain
8:05 Derivative
9:56 Integral
12:45 Why use integral
13:21 Tuning
16:15 Troubleshooting
These more updated tutorials really help as someone who just recently hit the 100h mark ingame
This video is amazing, thank you, ive been using PID for ages but when i turn off AI they stop working, i didnt realise there were GP PID's this is exactly what i was looking for, thank you
Thanks, these tutorials are helpful
Exactly what I needed, very well made tutorial video!
Amazing. Thanks for this.
Now I can finally get my airship off the water
One thing that this was lacking was saying what a fake set point is. Otherwise, this is a great and accurate tutorial.
thank you so much just picked it up recently this is a big help
@13:13 The point to which I will most likely be referring people.
eyy now i can link a vid instead of just telling somebody to scroll through your steam tutorial :D
Thanks man this will help alot with my attack heli I'm making
Things to note what i found up by tests and how i understand it:
GAIN - % of propulsion linked with PID it has AUTHORITY over, the higher the less force propulsion has in certain situations when PID try to smooth agile maneuever out.
DERIVATIVE - Good for tweaking, this "prediction" acts like buffer and accelerator - it will help with recoil if your ship wobble after fire and can help with to rapid as well as slugish movement - WARNING, high values can INCREASE power consumption, while giving no adventage, test not only in calm but movement scenarios and set it so ship regains balance fast.
INTEGRAL - Is a limiter that forces PID to apply it's fix in set period of time. High value gives more time for fix and is more fluid (large bombardment platforms) while low values force rapid fix (small agile gunships).
BASED UPON propulsion placement, ammounth and is it roll/pitch/yaw/hover/strafe PID values will differ horribly. Also using propulsion for both ...examp. Hover and Pitch...is questionable sometimes. Atmosphere and Space settings are hard to combine, also AI maneuever/Behaviour and center of mass will have their say, so every change on ship needs PID check.
ONCE AGAIN ...stabilizing ship in calm situation is not enough, if force/impulse is applied it need to regain balance, also unproper setting might give you "stabilization" in both situations while power consumption might be higer by 200% of what's actually needed !!! (What can looks like "stable" and "calm" visualy in game FX in reality in game code those engines can fire for example left/right constantly trying to even themself out).
A lot of people have their own theories about PIDs and are unfortunately wrong. I can't fault people, it's not information that is readily available and there's details I don't quite get even now. That said, a PID without a gain of 1 can use 100% of the available thrust depending on the difference between the setpoint and the current state of the vehicle. Integral also "accumulates" error over time rather than forcing the PID to fix the error within that time though I can see why you'd think that way given a PID with low integral will push quite hard quite quickly if it's struggling to stabilise the vehicle.
As for not allowing each thruster to control every possible axis it can control, it's a whole other debate and there's pros and cons to each approach. However, I am quite confident when I say that enabling more than one axis for most thrusters is the more efficient option in the end, but it does require careful tuning and to plan ahead for situations you might not even imagine before you see it happen.
If you're familiar with a spring-mass-damper system, essentially the proportional component is like a spring and the derivative component is like a damper. Here, the proportional gain is equivalent to the spring constant. In the form that's in the game, the damping coefficient is equivalent to the product of the proportional gain and the derivative time. I don't want to get into the math of the integral component, but just know that in the form present in the game its relative influence is proportional to the ratio of proportional gain to integral time. I would avoid using integral control unless you have steady-state error you need to get rid of. Besides eliminating steady-state error, when your set point changes integral control will decrease the time it takes to get there but will then cause overshoot due to the accumulation of error.
I'm aware that some of the specifics of how I explained things do not necessarily fully line up with the math. That said, I think it does a decent job of explaining it as simply as possible for the less math-inclined people.
As far as integral goes, yes, it can cause overshoot. However, for combat vehicles where thrust is likely to be damaged, integral is a way to maintain control for longer. For that reason, I actually encourage people to use it even if they don't need to.
You summed up what I would have commented otherwise! Accumulation of error on the I term is an issue in this game, which makes PD controllers (I set to 250s) a safer and just slightly less accurate fallback. I only ever use the I term on altitude control, and pitch for frontsiders that may need it.
Thanks man the first tutorial that helped!
No joke i came here to understand control diagrams for my engineering course better lol, ftd players know their way to make complex concepts like this seem easier
I did not expect this, but I'm glad I could help!
Your tutorials are really top notch, Ohm! No doubts they will be invaluable to newcomers, I know they would have saved me a lot of hassle back when I started playing this game. Do you plan to make a part 3 to the Breadboard series?
Thank you! I'm definitely considering more videos on bread in the future. :)
honestly one of the best tutorials I've ever seen, amazing!
My goodness thank you bro!
I've been attempting to make a hover ship but couldn't really figure it out when it came to pids to make it stable this video has been very helpful.
I've been looking for a tutorial that explained the equations variables instead of telling what they use on their particular craft. My thanks for the clarifications. I can do a lot more with an understood problem vs than a given answer.
thank you, this has helped a lot. (I'm the Dr.Doofus person from Borderwise's discord)
Oh my god I finally found a video that helped me understand how to tweak a PID! Thank you so much, the difference between my pid craft and the ones I see on UA-cam was driving me insane!
What the times "actually" mean, in as simple terms as I can make it, is: Integral looks at the average of the error over the last integral time seconds, and tries to correct that. Derivative predicts how big the error will be in derivative time seconds, and tries to correct that.
The more explanations the better! :P
You have the most nerdy name I´ve ever encountered, I like it.
Y O U ...should do Tutorials for FTD, and i think you know it too... and that's why official Plane tutorial got you so ..."passionate" =P
Thank you for your feelings. I don't think it's fair to say I should have the monopoly on tutorials. I am a trained teacher which helps, but FtD is complex and different people will prefer different approaches to explaining things. I'm not proud that I am so harsh sometimes when people mean well, but I am indeed passionate about educational material... sometimes it's good, sometimes I should take a chill pill. ¯\_(ツ)_/¯
Bro please update this on god i need this
This is great. Doesn't seem to work for boats though. At all. My yaw is all over the place.
It should work fine for ships, though the ratio between thrust-to-weight and water resistance will make tweaking things harder. Also, yaw tends to be more finicky just in general and I would definitely recommend against using integral for yaw.
@@OhmIsFutile Thank you very much. I didn't expect an answer to my vague complaining. And it's actually helpful ^_^