A Manifesto on the Relationship Between Fighters, Titans and AI

[SwankyCowboy]

Conclusion Two: Carrier-based respawn points are necessary to allow for PC-controlled fighters

Rather than having the player character physically aboard said fighter simply have them at a terminal on the carrier ship controlling the fighter remotely. That way when the small fighter inevitable gets destroyed the player could then take over control from one of the other AI controlled fighters.

 

[AsherMaximum]

We could also work around this by making undeathinators turn off while the ship is overheating.

Or, make undeathinators on a ship have a countdown before respawning a player.

 

[Izikiel]

Rather than having the player character physically aboard said fighter simply have them at a terminal on the carrier ship controlling the fighter remotely. That way when the small fighter inevitable gets destroyed the player could then take over control from one of the other AI controlled fighters.

That'd be kinda cool. In addition, we are already controlling them through one terminal, haha.

 

[Mariux]

Rather than having the player character physically aboard said fighter simply have them at a terminal on the carrier ship controlling the fighter remotely. That way when the small fighter inevitable gets destroyed the player could then take over control from one of the other AI controlled fighters.

I think we should have both options: AI drones and player controlled fighters. Except anything that's remotely controlled, incl. turrets, would drain more power from it's systemss. That would make it more balanced.

 

[Criss]

I think we should have both options: AI drones and player controlled fighters. Except anything that's remotely controlled, incl. turrets, would drain more power from it's systemss. That would make it more balanced.

Power shouldn't be the only thing effected. The accuracy of a turret controlled by a player would be higher than that of an AI. I say this because if I were to get into a turret that would otherwise be more accurate than me, what would be the point of me getting in it to begin with? If I get in a turret it has to be so that I can make a difference. Sure some people might suck at aiming but I think we could do with a limit to a turrets AI capabilities.

 

[NeonSturm]

But if it is too limited, you get Luke-Skywalkers blowing up Death-Stars by throwing infinite bombs onto it from a tiny invulnerable spot

 

[FourFire]

That scenario only occurs if the deathstar has insufficient shield capabilities to withstand the tiny bomb weapon which fits in the tiny "invulnerable" spot (I think you meant blind spot?)

either way I agree this is an interesting proposal, and it shouldn't be too hard to nerf AI to be worse than the average player exempting scenarios where hardware limitations come into play (A massive swarm of 500 fighters where each has 10% hitchance is superior than 50 players who can't hit anything because of lag)

 

[NeonSturm]

Yes, "blind" spot and it was (partially) a joke. (can't edit )

Sometimes I want turrets more accurate if the owning player is logged out, or the mother-ship docked to a base.

I think in offence players should have advantage, in defence turrets. But that will be hard to justify.
In case you wonder why I think so: it would buff bases vs ships. And because bases more expensive than your flagship are stupid, they should be buffed more when your flagship is docked there (to avoid troll-bombing). This is one possible solution I came to (without invulnerable faction home).

 

[starhaiden]

automatic turrets could have a computer that controls the turrets one computer block = to 500 blocks on turret 100% efficiency 1:1000 = 50% efficienty

 

[z1967]

Possible methods of being able to control drone fleets could easily vary and possibly even be implemented together.

Method 1: Full Autonomy

This method would have the user (drone master) input a set of instructions in the AI and set what happens when the instructions are completed (loop, stop, execute other instructions etc.). For example, you may set a fighter to fly and target the closest enemy ship (or biggest, smallest, targeted etc.) then after either dying or completing the task it may target a specific docking port on a mothership for redocking.

This allows a part of the next mentioned method, and I understand that something similar is being planned.

Method 2: Follow the leader

Method 2 allows users to denote a "leader" ship that the fleet will follow. Thrust limiting will be used to make it so that your leader fighter does not leave the bomber's behind. The user should be able to either pilot the leader ship or set an AI on instructions to pilot the ship. If AI is used for the leader, then the user should be able to set instructions such as forcing followers to break formation and use code on board, regroup and retreat, and assign new leader when destroyed.

Follow the leader will be the best option for those who need escorts or need to fly a lot of ships one place without having to pilot each and every one. Also could be used for parades and stuff.

Method 3: Targetting and spotting

This method is best explained as you tagging a target and the AI doing the instructed action to the target (shoot, mine, dock etc.). A computer could be used to control this on both ends (a transmitter and a receiver, possibly just one block but it toggles between the two). This allows you to have more control over your drones but less autonomy. This would be great to have since you can then set follow, fly, attack, and defend commands.

AI will be important, especially if/when it gets overhauled.

 

[starhaiden]

I have a idea of how to move all your ships a long distance without getting al make a giant ship with a big hanger then dock all your ships, this might take a while depending on how many ships you have. Then fly the ship to your destination

 

[Gerion]

I do like the economic site of balancing. With drydock/shipyard and a cargo system, fielding a titan in the first place would be a sign that you talk serious business, and loosing it would be a major setback for any faction. Contrary to that, fighters would be throw away things you hurl at your enemy in spades because rebuilding them is easy and inexpensive.

And this is were the actual balancing between ships should be in my opinion: A big ship with adequate point defense weapons should be a tough nut to crack for fighters and bombers, but remember the key thing: Every point defense weapon you put on your ship decreases the amount of anti-capital weapons you have on it, just because there is only so much space on the hull. So a cruiser decked out with anti-capital weaponry (unguided missiles/big cannons for example) would most likely best another cruiser that had more mixed weaponry and would definetly destroy one that is designed with pure point defense in mind. Of course, that cruiser would probably fall victim to fighters itself. Specialisation at work.

For this to work, dedicated point defense weapons need to be straight up better at destroying smaller targets while non-point-defense weapons need to be great against bigger targets without the two musling on eachothers turfs. One thing that springs to mind is the system from Masters of Orion 2 (great game, btw). You could directly assign weapons as point defense or heavy mount (was it called that?).

How that could work in Starmade (I am writing with turrets in mind, as we are talking AI against AI):

Every turret (not the weapons themselves) has to be assigned to be either heavy mount or point defense. The effect applies to all weapons in that turret.

Point defense increases projectile velocity (less effective evasive manouvers, pew pew), reload speed (you need to take down many targets fast), turret turn speed (keep tracking, those fighters are agile), missile lock on speed (less tracking, more shooting) and number of shots in shotgun weapons (more fire in the sky... space). This all comes at greatly reduced damage values per shot, still enough force to be dangerous against targets that are smaller than your ship (assumption being that turrets and their weaponry scales upwards with ship size), but low enough that its only a minor annoyance to anything around your own ship size or bigger.

Heavy mount, on the other hand, decreases reload speed (less pew, more boom), projectile velocity (you dont need speed, the target can barely turn), turret turn rate (see velocity) and lock on times (see velocity also) while boosting damage greatly. All of this should make sure that these turrets arent effective against larger numbers of smaller vessels, while giving smaller ships weapons that can actually hurt their bigger cousins.

Of coarse, a small ship will still be utterly wrecked if it gets hit by a heavy mount (lock on) missile, but the combination of low turn rate, reload speed and projectile velocity/lock on time should make sure that point defense would be more suitable in that situation. Something else i thought about was to have the range be a factor, shorter range for PD, longer range for HM. But then again, i dont even know if weapon range is currently a thing, it doesnt look like it is...

The last thing I am unsure about is what to do with a ships main weapons. Make them the same thing? A fighter with heavy mount (slower, more ponderous but with more boom in slower intervals) against a fighter of the same type with point defense (nimbler, faster firing, less boom)? Or just the jack-of-all-traits middleground weapons the current weaponry would turn into? Or the ability to assign the weapons on the main ship on their own?... I dont know, suggestions?

 

[NeonSturm]

I guess the problems peoples have with large guns are occasional hits.

My new idea: Make weapons weaker the longer they are. Make them stronger again the further their outputs are spread.

Make outputs closer to each other weaker than more isolated ones:

output weight = min(distance from closest other) / number outputs
normalize output weight ( sum(output weight) == 1 )
output damage = weapon damage / number outputs * output weight​

Allow longer weapons where shots likely will miss small targets to be stronger:

# x,y and z are intermediate results.
x & y = avg(distance from outputs from closest other)
z = x+y
(WIP) damage multiplier = x *z/weapon length <= 1 (???)
(WIP) damage multiplier = (x +y* number outputs) / weapon length <= 1 (???)​

For the peoples which are bad in math (or feeling a bit lazy like me):

Your weapon has a max z after which you have diminishing returns, but you can increase this max z by spreading your shots out.

If you have 3 outputs - 2 very close together, 1 apart from those, the close-together ones should be weighted 25..33%, the apart one 50..33%

More powerful guns (long z) have the damage distributed over (r^2 *PI) thus do damage^0.5 to fighters.​

Want to disagree?

Haven't even decided myself if this would be sufficient to protect fighters from large guns, but it is one thing to consider

Definitely not suggesting this until it is more fleshed out and I am convinced it works as intended.​

 

[Gerion]

But then again, we dont want big ships to be wrecked just because some fighters (or smaller ships in general) show up and it happens to have big guns. Occasional hits with big guns are imo working as intended. They provide some defence but get overwhelmed incredibly quickly.

 

[ltmauve]

Possible methods of being able to control drone fleets could easily vary and possibly even be implemented together.

Method 1: Full Autonomy

This method would have the user (drone master) input a set of instructions in the AI and set what happens when the instructions are completed (loop, stop, execute other instructions etc.). For example, you may set a fighter to fly and target the closest enemy ship (or biggest, smallest, targeted etc.) then after either dying or completing the task it may target a specific docking port on a mothership for redocking.

This allows a part of the next mentioned method, and I understand that something similar is being planned.

Method 2: Follow the leader

Method 2 allows users to denote a "leader" ship that the fleet will follow. Thrust limiting will be used to make it so that your leader fighter does not leave the bomber's behind. The user should be able to either pilot the leader ship or set an AI on instructions to pilot the ship. If AI is used for the leader, then the user should be able to set instructions such as forcing followers to break formation and use code on board, regroup and retreat, and assign new leader when destroyed.

Follow the leader will be the best option for those who need escorts or need to fly a lot of ships one place without having to pilot each and every one. Also could be used for parades and stuff.

Method 3: Targetting and spotting

This method is best explained as you tagging a target and the AI doing the instructed action to the target (shoot, mine, dock etc.). A computer could be used to control this on both ends (a transmitter and a receiver, possibly just one block but it toggles between the two). This allows you to have more control over your drones but less autonomy. This would be great to have since you can then set follow, fly, attack, and defend commands.

AI will be important, especially if/when it gets overhauled.

Target/spotting would probably be an override for normal behavior. You select an AI, and then tell it what to do. (also, if it's in an AI squad, the entire squad will do it.)

 

[starhaiden]

just build a battle ship with shield drainers and use logic to repeat fire it, point your ship at the enemy and fire your weapons

 

[Yetimania]

Holy cow! Just came back after a several month hiatus to find this thread stickied. Archive binging now...[DOUBLEPOST=1410163018,1410162897][/DOUBLEPOST]

I did SO enjoy reading your post, well thought out, and essentially a far better description of how I see things than i could have written out myself.
Thanks for writing this, and Stickied!

PS, I added "Titans" to your title, just for clarity on the thread page (you can undo that if you prefer)

Whoa, thanks! Also, you're right. That is a better title. My little post became more comprehensive and wide-reaching (and longer) as I wrote it than I had intended it to be.

 

[Esarai]

I love these suggestions! I absolutely love them, primarily because they give the dozen or so players you usually find in a faction the ability to stage large fleet battles, and make it so those battles aren't going to decided by the number and weight of ships fielded by the opponents. There'd be actual tactics and strategy to the battle.

One thing I think gets in the way of small-ship combat is that all missiles are inherently equal in the current game design. All missiles track all targets with equal effectiveness. Your planet-shattering ICBM can track a superfast, light fighter just as easily as a fighter-based missile could. And that leads to the situation where fighter pilots

can be taken out of a battle because a point defense turret so much as glances at you...

Well, not just point defense turrets. When a anti-capital torpedo battery can be used on a fighter just like a point-defense missile could be, there's no point in having fighters anymore, and nothing to keep someone from making EVERY turret they have an anti-capital launcher. I imagine fixing this by changing how missiles maneuver based on their damage potential. Big, planet-killing bombs should have a very large turn radius, making them ineffective against smaller combatants and necessitating shipbuilders to consider point-defense in-depth. Most space-sims you see out there have a mechanic like this so that each missile has a specific role with benefits and trade-offs.

But, like Yeti said... the AI is kinda... dumb. For turrets, this would be even worse. They have no target prioritization algorithms. They select and fire on targets at random. It's entirely possible that with scaled missile maneuverability, your anti-capital turret would start firing on a fighter, missing every shot, while your point defenses do negligible damage to the enemy titan bearing down on you. That would be a day-ruiner, wouldn't it? Sure you can set the big gun to 'fire on selected target,' but having to micromanage that during battle could easily become overwhelming. So give the AI the ability to prioritize a certain mass target based on the weapons installed on it. If the AI notices it's equipped with a planet-killer, it'll search for targets above 250k mass, and prioritize all targets in descending order of mass, heaviest first. If it's got a bunch of 1-block missiles strapped to it, 'ooh, I'm point defense, COME AT ME, FIGHTERBROS!' and it'll prioritize in ascending order of mass, targeting lightest before moving to heavier. It would also be nice to check if the turret can see the target, and have it reselect targets when it can no longer fire on its current target.

Again, I love Yeti's suggestions, and I think these would help realize the vision he's aiming for.

EDIT: Forgot to mention, I'm in favor of shifting balance towards mixed-weight fleets because it frees up game time. I'm part of the Tethys (soon to be Ragnarok) Galaxy RP server which does a lot of fun events, and recently we had a full-scale galactic war break out, but after that happened everything got quiet because pretty much everyone and their mum and their mum's dog skippy is building a battleship or dreadnought. When smaller ships can be effective, people will be content to enter battle with a smaller vessel and more events can happen more frequently, enriching most (if not all) RP environments.

 

[Yetimania]

When a anti-capital torpedo battery can be used on a fighter just like a point-defense missile could be, there's no point in having fighters anymore, and nothing to keep someone from making EVERY turret they have an anti-capital launcher.

The reason not do make every turret an anti-cap-ship turret is that anti-cap-ship-turrets have to be much, much bigger than point defence turrets. You could probably fit ten anti-fighter turrets in the same size as one anti cap-ship turret.

Anyway, thank you so much for what you said! That was really nice.

I absolutely agree with you, however, that you should be able to prioritize targets with missiles. Maybe an option in the weapons screen to say "aim for largest first" or "smallest first" or something. A similar option on BOBBY would be great.

 

[Gerion]

Make that a

you should be able to prioritize targets with all weapons

And then we are getting somewhere. And yes, you can use anti-dreadnought missiles against fighters, even destroy them utterly with big explosions, swarm missiles and lock ons. IF you hit, if they dont dodge, and if there arent so many fighters that you cant swat them all with your big, ponderous, slow flying, slow reloading, slow locking torpedos.