[chief1983]

There already are closer approximations of newtonian physics in the FS engine, but the one thing they all lack is the capability for the AI to use it properly.  What do you mean Why AI only?

[Topgun]

Why AI only?
Besides, that would ruin the gameplay, to say the least.
For it to work we would have to redo the engine and the conception of battles.

Want newtonian physics, play Frontier.

for mods man. for mods.
and I would be happy with semi-Newtonian physics. but that doesn't work good either. I tried it and the ai was way to stupid.

[karajorma]

Actually from what I understand the AI works perfectly well with Newtonian physics in The Minbari Project. And we have the code for that.

[chief1983]

and I would be happy with semi-Newtonian physics. but that doesn't work good well either.

Fixed.

[Topgun]

[Sushi]

Yeah, the AI does decently with newtonian and semi-newtonian, although some of the dodge & jink code doesn't work as well (the effects are reduced at high dampening levels). On the other hand, some of the features usually present with (semi) newtonian stuff (such as glide and sidethrust) don't really get used by the AI. You'll notice that those are the first two items on the list.

Also, you can probably get some benefit out of optimizing AI profiles for newtonian(ish) fighting. I'm sure the Minbari Project folks have done something along those lines, ask them about it maybe?

[haloboy100]

How about AI profiles? So that you can choose what kind of behavoir to flag on an AI when you create a ship; things such as "aggressive", "defensive" "conservative", etc...

That way some of these suggested features can be toggled on and off depending on how you want a certain A.I. to act.

Oh, and here's one. How about different A.I. behavoir during subspace missions? From what I've heard, the A.I. is easy to beat in subspace missions because they rely on shields (which are non-existant). So I think their tactics should change to be aware of no shields.

and perhaps squad meneuvers? Such as outnumbering an opponent (alpha one) and using certain flight tactics to take individual targets down. Such as, if it's one on three, one fighter would engage the target directly while the other two attempt to flank and attack simultaneously.

[General Battuta]

Hey, I've got a question about AI in general.

Do people generally experience a crash when the AI is ordered to fire Trebuchets at the player via good-secondary-time? I've had a mission crashing pretty consistently and that seems to be the cause.

[Bobboau]

well if this is a wishlist how about this.

intelligent object collision avoidance.

tell the AI to blow up the reactor, it finds a path through a bunch of obstacles and a few tunnels and blows it up and is able to find it's way out again.

[Stormkeeper]

I want my wingmen to actually be wingmen, not cannon-fodder pilots.

[General Battuta]

I want my wingmen to actually be wingmen, not cannon-fodder pilots.

As in covering you and all? Because other than their lack of specific formation flying tactics, I think they do a pretty good job already.

[aRaven]

the only way to prevent this is to have allied pilots have a  higher skill level than the opponents. like in XWA your wingmen rarely die, because they are on skilllevel superace.

[haloboy100]

the only way to prevent this is to have allied pilots have a  higher skill level than the opponents. like in XWA your wingmen rarely die, because they are on skilllevel superace.

really? I found that they always died.

And that they always, always missed their target whenever they were trying to attack the same TIE I was chasing...by a mile.

[aRaven]

you must be playing the original xwing then

[Trivial Psychic]

It might be possible to script in some wingman coverage (as in LUA), by having orders for, when in a 2-fighter formation, 2 automatically stays near 1 and only engages targets nearby.  If 1 is directly threatened by an enemy fighter, it will break off any engagements and assist.  1 would engage any enemy unless 2 gets into trouble.  In FS AI commands, assuming that Fighter 1 isn't the player, that would be:

Default
Fighter 1:  AI-Chase-Any 89
Fighter 2:  AI-Stay-Near-Ship 89
                 AI-Chase-Any 88

That should prevent Fighter 2 from straying too far from Fighter 1, while still engaging enemies.  The problem is finding the trigger for ordering one ship to break off from its current target, and protect its wingman.  Perhaps the same code trigger that orders the "cover me" message, or a check to register a certain number of shield impacts within a certain period of time, or combine that with a check to register the amount of registered damage to its shields.  Whatever the check the result would be (assuming that Fighter 1 got into trouble):

Protect
when
-<Fighter 1 is attacked chriteria>
-clear-goals
--Fighter 2
-add-goal
--Fighter 2
--ai-guard
---Fighter 1
---89

Now if there is a way within LUA to order Figther 2 to attack the ship that is actually causing the damage to Fighter 1, that would allow Fighter 2 to be returned to its default status if the attacker is destroyed.  There would probably need to be a check to make sure that the attacker is not a cap-ship, and perhaps to make sure that the attacker is within a certain distance to avoid having Fighter 2 drawn away to engage an enemy attacking Fighter 1 with long-range weaponry.  That would leave Fighter 1 exposed to additional attacks.  The chriteria to return to default orders, could instead of the above-mentioned is-attacker-destroyed check, could either be tempered by a don't-stray-from-fighter-1 check, or by a continuous check to see if Fighter 1 is still under attack.  If Fighter 1 is no longer in danger, then Fighter 2's default orders are reinstated.

There would also need to be an order sequence, so if say in a 4 fighter formation, where 1 and 2 are wingmen, and 3 and 4 are wingmen, and if Fighter 3 is lost, then 4 would join in to assist Fighter 2 to provide cover for Fighter 1.  If then Fighter 1 is lost, Fighter 2 becomes then new lead, while Fighter 4 becomes his cover... that sort of thing

Not even sure if that can be done, but its a start.

[Reprobator]

Hi, 'sorry for the necro ^^
I was tweaking around with turrets asteroids and other stuff and I've upgrade the asteroid's hitpoint by a factor of 100, the result is quite cool as now you can use them to hide and cover from enemy , the do not explode anymore with a single little laser shot...
But that's quite frustrating to see AI fighters wasting time and ammunition on those asteroids, and more than that, that's boring to see them collide with asteroids and stupidly explose while they try to destroy them ! (i vie upgrade the mass too)
So whats the point?
Well two thing in fact :
Having a flag in aiprofile for :
AI fighter/bomber ignore asteroids (they would just avoid colliding with them)
And have a new flag in the weapon.tbl like 'asteroid killer' that let cap ship with turret (armed with weapons using this flag) shooting those asteroids (with priority targeting orders like bombers/fighter/asteroids ).
That would be more realistic (IMHO) and i imagine being a gunner in those cap ship would get a bit more exciting than before

[TrashMan]

Regarding better capship AI manouvering:

The way I see it, something like this could work:


You pick one of the vectors that fits the ship best (and possibly a secondary vector if it's faster to get into that position) and put that vector in the ships table together with a defense vector
Forward, Back, Left, Right, Up and Down - combining those you can get all the vectors on the image.
It defaults to broadside. Now, sometimes you want some capital ships to be dumb, so maybe smart positioning could be turned on/off in FRED for individual capships.


On another note, I'd also like to see an aditional option - range. Basicly, what distance the ship will try to mantain from the target.

Long range would mean the ship would stay at the range of his (cap or supercap) weapon with longest range, pepper the enemy and fall back if necessary.
Medium would mean he would close to the distance where ALL of his anti-cap weapons would be in range. (take the effective range of all weapons capable of killing a capship, find the one with the shortest range and move to that distance)
Close is your regular bawl.

[General Battuta]

I just can't see a situation in which smarter capship AI wouldn't lead to ships ramming each other all over the place.

[chief1983]

They would of course also have to check the direction they want to move, lord knows most capital ships should have ample time to react to a possible collision.

[Sushi]

Stuff about asteroids

I agree that it would be nice if the AI worked better with high-hitpoint asteroids.

Regarding better capship AI manouvering:

Something like that could maybe work, but it would STILL be in every case inferior to scripting capship battles in FRED. AFAIK, the only thing you *can't* do for capships via FRED is control how much they roll. But that topic's already been debated to death in this thread: your ideas are definitely worthwhile if capship AI is ever improved.