Easy SubAvoidWeap Doctrine Fix and Maverick Missile Fix BETA

[LuftWolf]

Getting subs to avoid torpedos, especially the non-exploding on CM type from the LWAMI Mod, significantly more effectively is quite easy to accomplish.

If you change the line

LegCourse = ( TgtBrg + 150 + rnd 60 mod 360 )

to

LegCourse = ( TgtBrg + 100 + rnd 40 mod 360 )

the submarines will avoid torpedos by cruising at a course 100 to 140 degrees off the bearing of the incoming torpedo.

And, if you change

IF RND 10 > 2 THEN {
; Decoy or Jammer?
IF TgtSource $= "Active Intercept" OR ( rnd 100 > 65 ) THEN {
DEBUGOUT "CM Decoy"
CountermeasureIdx decoyIdx
} ELSE {
DEBUGOUT "CM Jammer"
CountermeasureIdx jammerIdx
} ENDIF
} ELSE {
DEBUGOUT "Decoy failed random draw."
} ENDIF

to

IF RND 10 > 0 THEN {
; Decoy or Jammer?
IF ( TgtSource $= "Active Intercept" ) OR ( rnd 100 > 65 ) THEN {
DEBUGOUT "CM Decoy"
CountermeasureIdx decoyIdx
} ELSE {
DEBUGOUT "CM Decoy II"
CountermeasureIdx decoyIdx ;LW always drop active CM
} ENDIF
} ELSE {
DEBUGOUT "Decoy failed random draw."
} ENDIF

then the submarine will always drop an active decoy before or during the evasion.

Quite simple but it is a major improvement in AI submarine torpedo evasion.

The only drawback to this is that the AI subs won't drop passive CM's anymore, but I don't ever use passive torpedos against subs anyway with the mod, as they aren't terribly effective, and active CM's actually do a fair job of spoofing passive torpedos anyway, so I really don't see any downsides.

This change turns any submarine into an Active CM dropping, datum clearing fool!

And it's so simple, you can make the changes yourself!

Although I'll send a version on to Bill once I fool with it some more so you can download it if you want. Once it is downloaded, you can add it to both your current Doctrine folder and your doctrine.LWAMI mod folder.

Or use it as a stand alone mod, if you're of that persuasion.

[LuftWolf]

Ok, I've sent this off to Bill.

Most nuke subs now will definately avoid most LWTs fired from medium range and the first pass of ADCAPs and UGSTs fired from around 5nm and a bit under, but if you guide them after the first pass, you still have a very good chance of a kill, as is even the case with most human players.

So, I can say, at minimum, the submarine weapon avoidance routines are essentially functional now with the non-CM exploding torpedos.

[Molon Labe]

I'm not sure active decoys are always right... and I think you have it the other way...passie decoys sometimes attract active torpedoes, I don't think it works the other way...

Is there a way to get the AI sub only to deploy passive if the torp isn't pinging, and to keep the torpedo out of its active intercept baffles to make sure it hears the pings and switches to the correct doctrine?

Come to think of it the change in the evasion course might be a sufficient fix to improve the AI's odds.

[LuftWolf]

Yeah, you're right, I got it backwards about the decoys... Although I don't think it matters for the time being. I'll explain.

My goal was to fix the doctrine without too much fuss, so getting the AI subs to drop passive or active decoys under certain conditions is a much more difficult problem, as the original doctrine tried to accomplish this, but it didn't seem to work out for whatever reason, and I didn't want to add another tier of programming language to get them to drop active and passive decoys at the same time, because then you'll get them firing way too MANY decoys within a given time frame, and since I'm not really a programmer, I'd probably introduce some kind of bug, without extensive trial and error.

In any case, given the very short passive seeker range and the need to launch passive torpedos at a slower speed to avoid sensor washout, the AI subs should easily be able to clear the datum given the new evasion geometry in the doctrine fix. As I said, I don't use passive torpedos against submarines, nor do I carry passive decoys on my sub anyway with the mod, although that is a matter of personal preference, I suppose.

This should be looked at as a quickfix, just changing what I had to work with in the most basic sense to produce the most dramatic change in effectiveness possible without too much fuss. Once Amizaur turns his full attention to this doctrine, I'm sure he can make it MUCH better. But in the mean time, this change is working very well so far.

[darksythe]

@LW Lets get together on thursday afternoon to test this new update... Gives a good excuse to get in a multistation game anyways. This time i promise i will not be drinking burbon before we play

[LuftWolf]

No problem.

Ok, I'll ICQ you sometime on Thursday.

[Deathblow]

Just made the changes just now. Good stuff.

[Molon Labe]

Originally Posted by LuftWolf

Yeah, you're right, I got it backwards about the decoys... Although I don't think it matters for the time being. I'll explain.

My goal was to fix the doctrine without too much fuss, so getting the AI subs to drop passive or active decoys under certain conditions is a much more difficult problem, as the original doctrine tried to accomplish this, but it didn't seem to work out for whatever reason, and I didn't want to add another tier of programming language to get them to drop active and passive decoys at the same time, because then you'll get them firing way too MANY decoys within a given time frame, and since I'm not really a programmer, I'd probably introduce some kind of bug, without extensive trial and error.

In any case, given the very short passive seeker range and the need to launch passive torpedos at a slower speed to avoid sensor washout, the AI subs should easily be able to clear the datum given the new evasion geometry in the doctrine fix. As I said, I don't use passive torpedos against submarines, nor do I carry passive decoys on my sub anyway with the mod, although that is a matter of personal preference, I suppose.

This should be looked at as a quickfix, just changing what I had to work with in the most basic sense to produce the most dramatic change in effectiveness possible without too much fuss. Once Amizaur turns his full attention to this doctrine, I'm sure he can make it MUCH better. But in the mean time, this change is working very well so far.

I think the reason the old doctrine wasn't working well was that the evasion course put the torpedo in the active baffles, so once the torp started pinging the AI sub didn't "know" to use active decoys. The hard work might already be done for you.

[LuftWolf]

Ok, I'll test that theory right now.

Give me a couple minutes.

[LuftWolf]

The AI is still dropping passive decoys when it would be best to drop actives. It's a boarderline case, but the penalty for dropping passive decoys when they should drop actives is much steeper than if they drop active decoys when they should drop passives, since they will usually easily clear the datum by the time a passive seeker can get in range of where they were.

With this database and the evasion geometry, the passive decoy won't really help the AI get away from passive torpedos, they are essentially redundant. However, in many situations where they drop a passive decoy when they should drop an active it is a fatal error.

So, if the passive decoys aren't really helping the AI subs evade the torpedos in a significant way, due to the fact that the seekers are really short range for quiet submarines and they clear the cone using their evasion patter, then there is no real downside to having the subs drop only active decoys.

This is based on a set of tests tonight only, but I highly doubt that passive torpedos will be made effective against submarines even if they never drop passive decoys.

[LuftWolf]

The essential issue is that, for the AI, the first turn and decoy drop are the crucial maneover for them to evade the torpedo. If they do that maneover correctly and drop the active decoy they can evade even very close shots, however, if they do the correct maneover and drop a passive decoy, their chances of survival decrease dramatically, even given the fact that sometimes the passive decoys will occasionally spoof an active torpedo.

In the same situation, for a passive torpedo, due to the combination of shorter seeker range, slow snaking pattern, and slow closure speed, the first evasion move alone is almost always good enough to evade the torpedo, even without a passive CM.

So, in my opinion, it is about making the AI take the correct action against the most dangerous threat to its survival, and that is an active torpedo attack at close range, and if it drops a passive decoy, as it too often does, then the game is up for that submarine in most cases.

Like I said, I've stopped carrying passive decoys on my dives, because I don't fear being sunk by a passive torpedo, so I think the AI captains would make the same decision.

[LuftWolf]

Yes, I just did a series of "worst case scenario" fires against an Akula using ADCAPs, and even under the most favorable conditions possible (firing head-on from a range of around 2nm), where passive decoys wouldn't even be of use, the ADCAPs have a lot of trouble keeping a track on the Akulas.

The torpedos may acquire, but they zig back and forth while tracking, and often actually start running parallel to the evading subs, but just out of the seeker cones, so that after a few minutes, the torpedos will actually overshoot the target slightly but still hold a tenuous track along side the sub. Then, when the sub slows down, the torpedos completely lose their track. In these tests, I had trouble getting the ADCAP to kill the submarine in basically any situation other than head-on under about 1.5nm, so that the torpedo acquired the Akula almost as soon as it came out the tube, and even then it was about 50-50 in killing it.

These tests have shown me conclusively that shooting passives against subs are a waste of expensive ordinance, carrying passive decoys are a waste of good Active decoy space, and that having AI subs drop passive decoys is totally reduntant.

Although, I'm still open to evidence that shows I'm incorrect in these conclusions... like if someone sank me with a passive torp in a MP dive, knowing that I carry no passive decoys. That might convince me.

[Amizaur]

Originally Posted by Molon Labe

I'm not sure active decoys are always right... and I think you have it the other way...passie decoys sometimes attract active torpedoes, I don't think it works the other way...

The reason probable is - active torpedos sometimes home with passive seeker, and then it's possible that passive decoy works against them :-).
Active CMs emit zero sound, this is set in database. Passive decoys have zero active sonar signature. So active decoy never can work against passive torp, and probably should a little because it makes noise when producing bubbles, also the cloud of bubbles could mask a target passive noise if it happened to be between.
The human player in most cases drops both active and passive decoys at first, because there may be passive torp after active one. AI sub should be no dumber, don't we want to make SP too easy ? AI inteligence is very limited anyway... I think it should relase both at start, and then "decide" what to "reload" next, so it may randomise passive and active then.
Other nice solution would be to add some passive noise to active decoy, and little active signature to passive decoy (10?). Then there probably would be some chance of active spoofing passive torp and a very little chance that very close passing active torp would be decoyed by a passive decoy. I believe that this was made in SCX but have to look there to confirm this.
The torpedo in active intercept baffles could be part of the problem, I seen funny behaviour when AI subs evaded sonobuoys - they turned away from they and run, but after some time they "forgot" that there was active threat behind them because the stopped to hear pings. So they slowed and turned, then heard pings again and everything started once again :-).
I think evade course values should exclude those that sets torp in AI baffles. But I understand that this is already done now ?
And the sub should not only run! When decoys don't detonate torps, this tactic is sucide. I've seen many AI subs running from active torp in stright line, dropping several active decoys and torpedo homed on some of them, but after passing through each decoy torpedo started to search again and the target sub was ditrectly in front of it. Boom. I think the sub should first drop a CM and run in the right direction as usuall, but when it drop SECOND CM while running fast, it should turn 90 degrees to not be in front of the torp when it passes through decoy. Just running straight and dropping decoys is very dumb currently :-/.

I didn't analyze evade scripts yet, but I wonder why there was a random, why could AI sub "want" to NOT relase active decoy if it was pinged by torp ? Maybe something like that - the condition to not relase could be only related to threat range, there could be a random chance that AI sub "knows" range to torpedo and then IF torpedo is far away, it would not relase active but only evade. If the roll was unfavourable, so AI sub doesn't "know" range to torp, it would always relase active decoy ? Would it be good tactic ? I'm not sure, I only show example of how this could be done, but what should be done in similar way (i.e. what tactic, what evade algorithm) should be determined by more experienced players.

[LuftWolf]

Bill has posted the updated SubAvoidWeap doctrine to his website.

Thanks Bill!

Enjoy everyone!

Cheers,
David

PS Don't forget to grab Amizaur's updated Maverick Missile Doctrine as well! Maybe Amizaur can post the link to it again here, as I can't seem to find it here on Subsim. :hmm:

[OneShot]

Even tho thats OT, but what was changed in the Maverick Doctrine?