boat keeps listing a lot even with a stabilizer

[DogboyGaming]

my boat that i am working on is a 1/2 scale Costa Concordia when i spawn it in it float fine but after i start driving it starts listing back and forth i think its too buoyant

[NotNice]

Its the Costa Concordia,so probably a realistic design feature for it to list to one side.

Apart from the jokes,the game's physics really do not like boats with tall superstructures/decks.Balancing a boat like that will take many weight blocks and fins.

[Siegfried67]

Originally posted by Hunter101:

Balancing a boat like that will take many weight blocks and fins.

No fins needed, and has been fixed without adding more weight blocks, just tunning the pid.
Of course, fins and more weight blocks will increase the stability even further, but it isn't required.
Just avoid having a 20 m tall ship with just 1 m draft.

[Jorg Hammond]

By the way, how did you figure out the PID settings, Siegfried67?
With the intuition coming from experience? Application of the bisection method? Mastery of black magick?

[Siegfried67]

I sacrificed a virgin to my dark lord...

No seriously, by looking at the counterweight moving.
It was going back to center position only once the ship was level. By the time the weight reached center positionl, the ship already tilted to other way. Happens when D is too low compared to P. Solution was to increase D or decrease P, and D was already at 20 000.
And in his setup, the counterweight was moving fast enough, P at 200 was acting on slider speed, so i guessed that decreasing to 100 should be enough, and worked.

Still capsized in storm, so maybe go a bit lower (80/90) and add some fixed weight to the bottom of the hull, the get it a bit lower in the water.

[Jorg Hammond]

It makes sense to me, thanks.

[DogboyGaming]

so its still going to rock a little one way or another right

[Ra-Ra-Rasputin]

PID can't really alone do for a balance thing, as it needs to automatically push toward the centerline as the ship starts straightening. Like in my extremely top-heavy thing i posted, the deeply simplified balance system returns the beam toward the middle so long that the delta of its tilting stays positive toward its centerline.
The returning is much slower than the actual balance adjustment, and if it ever returns too much, the electronics nudge it back until appropriate delta is reached.

[DogboyGaming]

ok

[Siegfried67]

Originally posted by Ra-Ra-Rasputin:

PID can't really alone do for a balance thing, as it needs to automatically push toward the centerline as the ship starts straightening. Like in my extremely top-heavy thing i posted, the deeply simplified balance system returns the beam toward the middle so long that the delta of its tilting stays positive toward its centerline.
The returning is much slower than the actual balance adjustment, and if it ever returns too much, the electronics nudge it back until appropriate delta is reached.

I don't use delta in my builds, but the tilt angle itself. Guess i'll give a shot with delta one day. But correct me if i missunderstood, with the tilt delta as PID input, if my boat is stable when tilting at 45° on side (imagine slight center of mass offset), wouldn't the PID just center the weight without correcting the tilt, delta being 0 ?

[Ra-Ra-Rasputin]

You need to account for both, and their relation for one-another.
At higher tilt, the positive delta towards the tilt 0 position needs to be higher, however, the arm should retract toward center line at all times to prevent overcorrection as the speed at which the slider can move is limited.

The way a simplified version of this works is that there's 2 parts of it fighting. One part watches the tilt, and tries to nudge the arm back to the center. The other, stronger part keeps an eye on the delta of the change. If it's either negative or not positive enough, it should kick the track in a desireable direction, overriding anything the tilt watcher is doing. You can examine this basic version in the last workshop creation i put up in this thread.

It's very basic for 2 reasons:
Reason 1 is i don't upload my more complex microcontrollers, because i've put in a plenty of work and consider them my little babies.
Reason 2 is that the former is practically impossible to understand because they're large and very complex. It's much easier for everyone if i strip most features and put in as-simple-as-possible variant up, so people who can look at it learn something.

[Siegfried67]

Originally posted by Ra-Ra-Rasputin:

You need to account for both, and their relation for one-another.
At higher tilt, the positive delta towards the tilt 0 position needs to be higher, however, the arm should retract toward center line at all times to prevent overcorrection as the speed at which the slider can move is limited.

The way a simplified version of this works is that there's 2 parts of it fighting. One part watches the tilt, and tries to nudge the arm back to the center. The other, stronger part keeps an eye on the delta of the change. If it's either negative or not positive enough, it should kick the track in a desireable direction, overriding anything the tilt watcher is doing. You can examine this basic version in the last workshop creation i put up in this thread.

It's very basic for 2 reasons:
Reason 1 is i don't upload my more complex microcontrollers, because i've put in a plenty of work and consider them my little babies.
Reason 2 is that the former is practically impossible to understand because they're large and very complex. It's much easier for everyone if i strip most features and put in as-simple-as-possible variant up, so people who can look at it learn something.

I just use the D to avoid overcorrection. Maybe your way gets better results, but for me it's good enough. A want my boats to tilt a bit to look more "boatlike" ^^. I'm satisfied with 1/200 of a turn in calm seas.
Don't get me wrong, i totally trust you on the technical things ;-)