Help with AI changes; aircraft not working right

[JadeKaiser]

So I stopped playing for a little while, maybe half a year or so? Something like that. Anyways, coming back to the game, for some reason I can't get any aircraft to work anymore.

AI PIDs were integrated into the AI, okay great. Better than great, that's wonderful!

But why does that make it so that now, "Point at and Maintain Distance" apparently ignores the distance it's set to maintain and tries to hump the air directly over the target when the Hover maneuver is used (and only the Hover maneuver... you know, the one maneuver that's actually good for Point and Maintain aircraft)? Why is it that airplanes which previously worked, now believe that flying in circles for all eternity is more important that anything else in the world?

I've checked and checked everything I can think of. I've gone over the PID settings, the thruster configs, the center of gravity, everything. As far as I can tell, there is not a single reason why either of these things should be happening. On my new craft, or on the old ones. But they are, and something has clearly changed and/or broken. I can't find anybody else talking about it to get a clue from about what I'm missing, and I could really use the help.

[Javelin99]

Alright.


First things first, if it's old files, don't use them. Without thoroughly gutting and replacing the innards, they have so many leftovers it can't recognize what the hell it is in the first or get fought like a general PID.

Second: Point at and maintain distance was actually for six axis and appears reconfigured for hover. My zeppelins didn't work whatsoever for hover despite having more than six axis, but worked on six axis. Some behaviors work entirely separately based on type of craft it is and it's adjustment settings than the names prescribe. If it is a normal plane, I have to ask if you are mad or something because they're not meant to have that in them at all. Instead it should be plane 2.0 or the original, and either ram [which if giving a high disengage, works for interceptors, fly over is for bombers, U-turn like flyover works with dive bombers and german tactic styled dive-fighters.]

If it is more styled after a space plane, I.E, if you made a ballistic missile or it uses thrusters or dediblades for yaw, hover, etc, the only applicable settings may vary. I.E, my cobra nuke has plane 2.0, and uses ram with no disengage to kamikaze nuke into a target, and has overdrived/ovverclocked thrustering to yaw to dramatically it appears to be slithering to avoid incoming fire.

If you're ripping off contraptions like the coffin nail, a zeppelin, or a spaceship, the only correct AI is six axis, and you reconfigure any extra drives via ACBs to tell the AI it can actually use them. [As it does send commands out using either, which aren't meant for them. I.E, it may end up too rolled in damage, and thus appears like it suffered a personality change and ties using pitch thrusters as yaw. Which enabling banking turns actually lets it be competent if using that type of AI instead. for hover, some commands are straight up missing. or you have to individually modify everything. Such as hover will not actually hover. Like dediblades, it'll only ever go the direction you set it as if at all [otherwise up.] And will need two ACBs at minimum to dictate it on hover command being able to go up and down. A third for none if you don't want a constant.

If you need PID fixing, purely modify integral only in it's own settings, and nothing else.
If it's a craft meant for high stability one way or another, and you don't care about loss of functionality, then modify with set points and integral.

If problems persist, it isn't an AI problem.


I.E for a plane, if it continually turns forever but it's next to a resource point, it has adopted a mentality akin to a gatherer. It's actually just circling in protection of the ring itself which is a standard procedure. If it does it out in the middle or nowhere, or out of control appearing when in combat, it's either overyawing, or it can't yaw enough. In that latter case, this literally means whatever you are using to move it forwards is forcing it to yaw. Get rid of it or lessen the power ratio.

If it over pitches, it needs a integral fix and more pitching elements. [Except thrusters, whereupon their default and only setting is being so powerful that relying on thrusters this does nothing. Instead you need one up and down thruster forwards and backwards of the design to steady it. Nothing else matters.]

If it over rolls, you need more rollers.

If it still rolls, you probably have too high a degree to banking turns.

If it's again like a coffin nail design and you're using these settings which have their own equivalents such as how much roll and pitch for movement, adjust until it stops looking like a drunken Beluga that just stole seventeen pounds of hardcore abstract.

Problem finally solved.

[JadeKaiser]

Originally posted by Javelin99:

Problem finally solved.

A lot of your advice is somewhat hard to understand, because you wrote it out while seeming to be confused about something I said. I referred to two entirely separate craft in my post. One was a brand new helicopter design, the other was a jet that I had already made.

The helicopter works fine using Hover movement, except solely when I give it the Point and Maintain Distance behavior, at which point it completely ignores the distance that it's supposed to maintain and just flies and stays directly above the target. Same thing when I set it up with Six Axis movement, if I'm remembering correctly, except that Six Axis was having trouble with basic maneuvers and staying in the air as well.

The plane doesn't work at all for movement, and just flies in circles. Attempting to make a new plane from scratch resulted in the same thing. And yes, the planes were indeed using airplane movement. I tried both types.

So... no. Problem not solved. You've recommended nothing that I didn't try, as far as I can tell. Though that may be more because you mixed some things up about the original question, rather than you not having the answer I need.

[Javelin99]

That's weird.

For the helicopter in hover, which I think that's the only way to get it working since I can't get it functioning like one any other way, all I can present would be asking if the preset for maintaining distance while facing the target is active. This is because behaviors don't always automatically activate like how the actual AI type can.

Plane is still the same as presented.
The only thing I can add would be asking if you use custom control surfacing or the original unmoving air parts for the plane. I guess I'll list it out as a list of questions:

1:If using a custom yawing fin, is the custom pivot tilted so the arrow is turned over facing the left side, and forwards or the opposing side? This matters because if placed wrong the controls are reversed. It also doesn't respond too well if the turn time is higher than 0.8 seconds from one end to the other.

2: If using the original fins, it seems like a dumb question but I do have to ask if the yawing fin has at least one part towards center of mass?

3: If using custom roll fins/ailerons, are they on the tips of the wing back, or sides? the fins can be placed on the body and extend out to the end of the wing, with the frontal actual wings should be, on either side of the craft, set to a positive 1 to lift properly. This however doesn't solve the issue of yawing, but it is more so a movement aid.

4: If using the originals, is the ailerons instead on the side-tips of the wings with the flat section facing forwards and the piece is upright? There should probably be about 3 - 5 depending on how heavy it is.

5: Is banking turns enabled, and if using custom fins, are the pitcher pivots actually with the arrow upright facing the front? If the original, while they do work regardless and irrelevant, should be placed on either end of a tailplane because it gives better reference. Should be shorter in width than the wings.

6: If the above still doesn't end up working, you might have to replace yawing fins with either a jet, or for more refinement use some ions. Those tend to help with yawing issues without sacrificing the design of the craft in question. But it's now barely a plane as conventionally thought as.

If that still doesn't work you'll have to put up a model on the workshop so it can be examined. That would help.

[JadeKaiser]

For the heli, yeah, the Point and Maintain preset was turned on. In fact, it only ever went screwy when it was turned on. I tried other behaviors and they worked fine, it was just that one which seemed to break it.

Also, you can actually make Ship or Tank movement work for a heli or airship just fine as well, if you toggle on the "allow use of Hover axis" option in its settings. I can't remember if I tried that for the heli before, but I've used it in other cases.

As a farther mystery, I just went to test Ship or Tank on the heli while I was typing this up, and both it and Hover movement are working fine with the Point and Maintain behavior now. Which really just deepens the mystery, because there hasn't been an update and I actually took the time to verify local game files when I couldn't get it to work before, to no effect.

For the plane, I tried a few different maneuvering setups when I built the new one from scratch, and they all had the same problem. Thrusters, control surfaces, etc. The older design was one I actually posted on the workshop when I first made it, just a setup where I made something decent using the fighter jet prefab hull.
https://steamcommunity.com/sharedfiles/filedetails/?id=2470247554
That's the workshop link. The version posted is from before I started messing around with it to try and fix things, but nothing I did really made a difference anyways. Just to be sure, since the heli is now inexplicably working I also tested the jet again. Still broken.

As another note, I also made a prop plane when the jets weren't working, and it did not give any of the same problems. Which makes things even more confusing.

[JadeKaiser]

The thing with the jets is really, really weird honestly. As far as I can tell, it's perfectly capable of moving exactly how I want it to be doing. It just... refuses to do anything but fly in circles, for no apparent reason. Tested both with and without banking turns, checked to make sure all control surfaces were correctly oriented, etc. etc.

[Javelin99]

I'll make a follow up post or edit this one after I've looked at it, I am going in now since I am half done with my kraken build.

[Javelin99]

From what I am seeing I was correct with the initial assessment. Something onboard is forcing it's yaw to be so extreme that nothing on board the craft itself can correct it. I fixed it by adding three jet thrusters to either side at the front. That allows it to flyover the target. Also unless I messed something up while looking through it, one of the custom jets settings is placed on 1 for forwards, while the other is something like 0.95. The settings enabled to help with AI also only has "fly" enabled on it.
[Edit]
I enabled the extra settings, and I rebuilt the engine exactly as you had it set up because on spawn the bloody thing kept complaining nonsensically that propulsion was somehow backwards when none of it was. I haven't gotten an issue with it after.

[JadeKaiser]

Okay, so I took a look at it again myself using what you just said, and I just don't see it. The stuff with the engines being unequal? No, they're the same.

The troubleshooter settings shouldn't change anything, and none of them besides "fly" actually apply because it *can't* do those things. It doesn't have anything that responds to the Hover axis, or anything that should be doing so, because it keeps in the air via thrust and pitch control alone. It can't remain stationary, strafe, reverse, or turn on the spot. The only one left, the only one that applies is "fly." Those don't actually change anything in the AI's behavior anyways; they're just a checklist for the ease of you or I, the builders' benefit so we can see what might be causing problems.

Adding the sideways-facing jets to the front does seem to let it go where it's supposed to, as you say, but is immensely frustrating as a "solution." I would much rather like to find and fix the actual problem than rely on brute-forcing it with wasted cost and engine power. At this point, I'm starting to think the problem is that there's some kind of bug in the current version of FtD relating to thrust and yaw control surfaces at the back of the vehicle.

The warning about reversed propulsion is a bug that has always been there. Seems it finally got fixed to where rebuilding the jets removes it now, but it hasn't caused issues before and fixing it doesn't seem to be eliminating any issues now. Only thing I can think of that could have been causing it is some of the CJE components that don't *actually* care about direction being turned "backwards" or something, and I had tested for that when I first made it to no result. I basically just gave up on it at the time, because nothing was fixing it and it wasn't doing any harm.

Fake!EDIT: Okay, I just tried swapping out the Aero Rudders for custom control surfaces, and I found something interesting. Custom control surfaces show very clearly how they're trying to turn, unlike the aero rudders, which is how I was able to see it.

The custom control surfaces set to Yaw will slowly turn so they are actively turning it to the right, stay there for a few seconds, then snap over to full left to correct it and immediately drift back to push right. This would seem to indicate something screwy with the PID, except that the jets on the front are NOT having the same problem despite listening to the same axis. Also because I tried working with the PID before to no avail, but if it weren't for the jets then someone could dismiss it as me being bad with PIDs. If it was actually a problem with how I set up my AI, the jets would be screwed up in the same way.

From this we can conclude that there is DEFINITELY a bug involved.

[Javelin99]

Things like the control altitude element for airplanes and all isn't strict to that however. But alright.

This is also a custom jets issue, which may be where the bug is.
Despite being at the center of mass or around, I have found that designs which have them too close together if more than one at center tends to generate intense yawing problems. This may also be due to their intense speeds, the drag simulated, as well as thrust generated for the speed, despite being effectively a straight line for designs like this one. And there isn't a setting enabled where the main propulsion can also be utilized to help with yaw. [Which for a double jet, would be an interesting concept.]

It is akin to making missile systems where the faster the thing moves, the more fins or proper turning thrusters to get it to yaw properly. In FTD there tends to also still be a thing from years old from ships, where improperly placed or too weak rudders cannot counteract yawing forces, which can be generated by the shape of craft, all the movements made [such as rolling intensively and pitching during this.] as well as the control surfaces used and their sizes/amounts. When I made an ICBM 42 meters long, 3 wide and tall, with a single custom jet inside of it, it didn't initially have any yawing problems. Instead, the intensified thrust/speed generated caused the craft to pitch. This also can be experienced with airships, submarines, ships, spacecraft [except high enough.] etc. So using this experience, we can theorize while it might still be a type of bug, it is not as conventionally thought.

So the original idea of testing the design with weaker engines like the standard jets, ions, or dediblades could also help with gathering data to confirm if it is an issue fully. [I.E, for some of my designs, yawing and pitching, sometimes rolling issues done by thrusters on too light or small of a craft was solved through intentional weakening.] Therefore another solution to this potential issue, would be limiting overall thrust or speed the custom jets can do and experiment if it works. As mentioned, weight also affects it greatly. [Whereas size is more so because there isn't room to put more in.] Whereas pitching, rolling, and diagonals are extremely easy to do on even the most lumbering of behemoths and require very little thrusters, blades, etc, the issues are always yawing and movement. [Sometimes the craft in question requires forwards movement to enable these, or rolling and pitch is used to replace yawing as a whole.] In terms of weight, if you have too powerful an engine, with intense speed ratings, and weight cannot be evened via size such as wanting a particular look, shape, or size, then pitching will become an issue. [For aircraft it's actually helpful though.] If we finally thus come to the twinjet of this jet fighter is, we start seeing a yawing thrust issue. [As with my nuke design I call a cobra, which intentionally does a slithering motion to avoid projectiles to then ram a nuke into people.]

So with this segway we can conclude with this:

I believe I see more of a thrust, control surface, as well as weight and placement issue. This may be worth investigating one more time.

So we start with 1: the thrust: The craft is a twin jet with them to either side rather than vertically, which provides intensified speed from thrust generated. Being not too far away from centerline to the sides, but too close to it, this can affect stability. This also would cause a varied but slighter yawing issue. [Whereas if it was placed two meters to the side of the current position, the yaw would be more violent I believe.] So to solve this problem, experimenting with speed/thrust settings could potentially weaken it enough for the existing rudders on board to take control. [We will return to this later and why I think it is a concern.]

2: Weight. The craft the devs added in that you used is light alloys, which is great for floatation and speed. It has some armor, yada yada ya. But it has an issue: It is a horrible material as it currently is, because of the fact that it varies still from single blocks [like those near where the yaw jets if built again would be.], all the way to the better 3 meter beams. Some 4. Most of the craft is composed of singular to twos. This means that it's extremely light. However with custom jets, I've made heavy armor jets with intense pitch issues with just one large one which still threw me violently at 155 meters per second with an acceleration of 3 seconds. This leads me to believe the prefab actually wasn't made for them despite looking perfect for the job. It may of been actually for the original normal jets, which considering it is light alloy but large, would still give it a high speed but insanely weaker. This could also solve the problem. If it was heavy armor it'd be slower but more stable. If it was normal metal it would still be too small but better armored.

3:Placement: Because of previously mentioned issues, there is a connector type for exhausts, I think one is a corner pipe. You could try moving the thrust director a meter to the side from current placement and see if it more properly lines up and lessens the yawing force. [As again, the only issue the craft really has is that.]

4: Control surface failure.
This is simple and I noticed it the moment I got it on. Since I also experimented once with the hull but with augmentations a week ago, I noticed this: While the design can have more than enough for rollers, pitchers including the jets [the one reason why the custom jet design isn't violently blasting off to space to begin with just a tiny bit.], it's yawing fin ability, within the armored section likes yours, is exceedingly weak and brought me to the conclusion that it wasn't meant for this type of engine. Even when all pieces are removed beyond the "corner" wedging to keep protection and nice shape, there simply isn't enough rudders on either side to counteract the custom jets to either side. This is why even with custom control surfacing, the turn response is garbage, but the moment turning jets are put on, it works normally. There isn't enough yawing force to counteract the natural turning forces of the craft.

Even replacing the entire tailfin armor but not expanding beyond it would do nothing. The jets are too strong, and the craft itself too wide for it to make any difference at all. In order to create enough yawing force in order to counteract how it currently turns, you'd have to sacrifice most of the entire surface area. Again going back to how on the opening ends on the back, how it most likely was meant for the one piece jets.

As for the also noticed custom control wings popping a jig, that itself isn't actually a bug. It happens when the AI becomes in a confused state and then refuses to use control surfaces for intended purposes but instead will use them to attempt stabilization. Which in rare cases, just like if a control surface was too weak to be noticeable, placed wrong, missing, etc, could result inevitably in a "AI cannot control itself, see settings" error. Which despite the rudders being so weak to be useless, they apparently provide enough force that this message never appears. They can respond fast enough, do enough force, and etc for this to be prevented, but still appear to do nothing whatsoever. Therefore, a true solution would be if wanting to keep this design as is, would be either replacing the entire rudder system with jets like mentioned, utilize a smaller or weaker jet or entirely different type, if the placement solution actually worked, or finally to give it such large weight that it becomes slow and weak enough to allow current or expanded control surfaces to gain control.

There's also still a number of open spaces within the body of the jet itself entirely unused I believe, which can also be used to insert more rudders.

[JadeKaiser]

The problem actually does not seem to be that the rudders can't do anything. The rudders *are* doing things. In fact, them doing things seems to be the problem. Because what they are doing is exactly something that would cause the problem I am experiencing. When they briefly try to turn the plane left, it *does* start to turn left. For the quarter-second that it takes them to drift back to neutral, and then another quarter-second to reach full-right. They are absolutely capable of turning it where it needs to go, they just don't.

Also, the control surfaces not having enough yawing force due to too much thrust is ridiculous. The amount of yawing force they exert scales linearly based on the speed of the craft. If the thrust force increases, so too does the yawing force of the control surfaces. It's literally impossible for greater thrust to make them exert too *little* turning power where less thrust will avoid such. The opposite could possibly occur, but never that.

What I could see, is the game's programming for them failing to *recognize* the amount of turning power they can exert at that strength, and getting confused and causing problems as a result. However, this would have to be something that only recently came to be a thing, because the plane used to work fine. Not "used to" as in "several years ago." "Used to" as in "last mainline update."

[Javelin99]

Bro your Jet is kinda weird but I tried another technique because I was curious of if that would of helped. So I tried all my suggestions and besides minor to nothing, they appeared irrelevant beyond making the frequency of error messages look as frequent as if my screen was warning me of being locked on by a heat seeking missile.

I had to respawn and despawn it a few times to revert changes, and do one tiny thing at a time to make certain I did it right and was reproducable, but this is the closest standalone I got left.

So I went to look up on the PIDs

So basically I finally respawned it twice after the others. And what I did was on the one I left the fins as was but modified the PIDs to "nearly" defaults, and another I added more default control rudders as was already on there, but replaced all the wedge backs in the tail-fin armor section.

What I found was when I made the PID in the mainframe have the settings of a 250 integral, and a hard 1 on gain and deriv, it started trying to face the target, but the roll was too steep and it kept trying to circle.

So on the final iteration where I added extra rudders and replicated my settings, it started flying over unsteadily the DWG crossbones over and over again, but sometimes does a roll but will pitch into the target direction. The roll making it unsteady isn't all that bad [and I am certain messing with the PID can solve this, that is what I did with my rubber dolphin.], and as it is closer to the target, it starts to steady more with the rolling becoming less noticeable. Then it becomes more wacky as it gets further away and tries to come back in again.

So the problem actually seems to have yet again been, a yawing issue, but also a PID setting problem. The issue here however isn't even yours. It would appear the PIDs are just jacked up. [And I don't think it needs as many rudders as I put on there, but I gave it the entire thing but the corners and the one top block just to be certain.] The only issue that is with the other would be it wouldn't have the means to turn in, and would struggle. Other times it'd entirely avoid going into the target direction for a flyover and would revert to circling behaviors. So to counteract that, I thus did the above.

[Clarification edit] All I modified by the way was the yawing PID, but left everything else as is. Unless somehow modifying one mods them all, shouldn't have too many changes.

[JadeKaiser]

So end conclusion roughly the same as mine, then? That there's something buggy with PIDs and their use of yaw control surfaces right now?

[Javelin99]

Originally posted by JadeKaiser:

So end conclusion roughly the same as mine, then? That there's something buggy with PIDs and their use of yaw control surfaces right now?

Pretty much equalized. The PIDs with Yaw pretty much seem to suck and depending on the settings would send sudden spikes breaking it and trying to viciously yank the rudders around or making them stay still when they shouldn't, and then plane itself needed another set of rudders to turn once the PID issue was corrected to enable it to complete a turn for flyby. For some reason it still gets "AI can't this view settings" even though everything about it works just fine. I can't seem to find what the problem with that is, so like some of my airships I'll just ignore that since nothing fundamental is broken or declining.

I think it is safe to say that until fixed, PIDs beyond dramatically editing roll changes should probably be avoided. Though Pitch might be fine too depending on how powerful the pitch is and if the PID can even recognize it properly. I am unsure as to why Yaw is more of a mistake than adopting a redheaded step child named Crognak the Yeetinator.

[JadeKaiser]

Well, based on the fact that adding the jet thrusters was able to overpower the problem, and those were also being controlled by the PID, I don't think that PIDs necessarily need to be avoided. Rather, it's using them with control surfaces specifically that causes a problem. Which is not ideal at all, but at least it's something. Something pretty important, because trying to make a plane that will fly without the fine-tuning a PID provides is an exercise in futility.

Though even then, I wonder why my prop plane still worked fine; it's not like I wasn't using aero rudders for that thing's yaw too.

[Javelin99]

Originally posted by JadeKaiser:

Something pretty important, because trying to make a plane that will fly without the fine-tuning a PID provides is an exercise in futility.

Though even then, I wonder why my prop plane still worked fine; it's not like I wasn't using aero rudders for that thing's yaw too.

It's a little bit easier for prop planes though. Whereas with thrusters, one may have also noticed that even with the jets working on the side, they had to remain active, and they still kept a notable sway to try to keep it on point to the target.

With the prop plane all the force is only forwards, and there's not many that can go wrong with it. The moment part of the design is messed up and causes instability, slight marginal edits to the mainframe PID, or sometimes behavior and AI settings would fix it. But they move slower. [Though you can also do dumber things with them.]

Jets can be larger and more fun at times though, because you could use the overpowering of the thrusters to an advantage. Like kamikaze tactical nukes, which sway around at intense speeds, or a heavily armored flying squirrel, or distraction support craft filled with heat and other aspects to divert fire from a more important vessel while also preventing its own destruction.

They more often need less PID tuning and at times none at all than jets do. I often find the only settings I need altered that isn't working for me or envisioned is roll.

I think Submarine tuning is more apocalyptic though. But I stopped bothering with that until recently when I just used ACBs for a year on end to make it move around, avoid terrain, etc. I think later I might try to see if an ACB-brained craft could be workable. But would probably throw a mainframe in there somewhere.