Infantry TAC AI - trying not to rant

[A Canadian Cat]

If you watch footage from helmet cams etc in Afghanistan, there is a heck of a lot of hunkering down in wadis when rounds are incoming while everyone tries to work out where the hell it is coming from (no laser like tracers here).  I wonder if "cowering" is an unfortunate expression in game?  Is it possible for a squad to become pinned by volume of incoming fire without there being an associated morale drop?  Or does a drop in morale precede the cowering?

Oh yeah, cowering is not "I want to go home to mommy" its "crap that was close stay down". Perfectly OK and well motivated guys cower when the bullets start flying. At least that's how I look at it anyway.

[Guest]

Ian's understanding of "cowering" is correct - it's more widely applicable as a "wow, that was close" sort of response.

 

If you take fire, how do you decide what to do?  What are the variables that factor into it?  Distance, effectiveness of your weapons systems, the enemy weapons systems, fatigue, morale, terrain, known enemy / friendly positions, whether you're already taking fire from somewhere else?  Specifically, if you're moving, what would lead you to alternately:

 

a ) Continue advancing.

b ) Find hard cover and move towards it instead.

c ) Hunker down and remain in place.

 

Also, under what conditions would you try to engage / suppress the enemy?

 

As a player, too, under what conditions would you like to NOT have control over your unit(s)?  If units are doing anything but a ), they're not doing what you told them to do.  Where's the line?

 

There are a lot of variables.  A significant amount of the time, I would guess that you'd want to break LOS with the enemy, regroup, and formulate a new plan of attack.  The quickest way to do this is to stop in place and head for the nearest concealment - the ground.  Clearly that's not ideal in a number of situations, but as a general response you can see why we went that way.

 

You can also hopefully see that any attempt at comprehensively addressing this would be a pretty huge endeavor, as all of the above sets of variables and responses would need to be coded, tested - and then probably still be "wrong" a fair amount of the time.  We'd be adding a whole new section of intelligence to the tactical AI - and WeGo players wouldn't be able to stop it once it had started, so factor that in as well.

 

If there were some very specific case that could be addressed - say, the *worst* example of general TacAI behavior - I'd suggest that we push the conversation in that direction.  Small, specific things can be fixed.  "Units don't respond well to fire" is going to be too big to tackle all at once.

[kulik]

Generally, I would like to see more return fire even from covering and pinned troops, however, less effective in terms of accuracy.

 

Other than that, troops seem to shoot only if they have a clear ID and visual contact with the enemy.

I would like to see tac ai firing on suspected enemy positions, and firing in general direction of the enemy in short range fights such as woods and buildings; or, if they can see a friendly squad being shoot at, they could also shoot at least in direction of their attacker.

[Melchior]

For every player who wants that kind of behavior their will be another who hates that his men are apt to give away their positions. The tenacity of your men to engage targets real or imagined should be a function of training and experience. Would I expect Green troops to light up the first passing donkey that looks like Fritz? Sure. Veterans and Crack not so much. 

 

If you've got a suspected enemy position you want fire on, "Target /Light".

[kulik]

More often than not, I would prefer them to be more trigger happy. For ambushing troops, target arcs are necessity even at present.

[Combatintman]

I seem to recall from basic training (1984) - the reaction to effective enemy fire drill was dash - down - crawl - observe - sights - fire.

[Wiggum15]

Generally, the first thing to do when someone is shooting at you is to shoot back. Thats something the TAC AI does not do very well.

They go down, crawl, take cover...but it takes too long till they return fire at known or suspected enemy positions.

[Pete Wenman]

 

I seem to recall from basic training (1984) - the reaction to effective enemy fire drill was dash - down - crawl - observe - sights - fire

This one Carl ?

http://community.battlefront.com/topic/118052-prioritizing-the-scenario-editor/?p=1580979

p

Edited February 14, 2015 by Pete Wenman

[vincere]

I seem to recall from basic training (1984) - the reaction to effective enemy fire drill was dash - down - crawl - observe - sights - fire.

 

It is, as is adjusting position after a few shots before firing again.

 

Often doesn't happen, even with trained and experienced soldiers. Both are tiring; and easy to forget when pumped.

[Sulman]

One thing I found in the new black sea is that the US army troops in particular seem to be extremely prone to being pinned and cowering. One or two shots from an AK will send an entire rifle squad into laying down and being shot one by one, even when surrounded by AFVs and having good command lines etc.

It's extremely unrealistic. US soldiers and marines would only devolve to such behavior under the most extreme conditions (continuous artillery suppression or continuous supression by a MG etc). It seems like the us army guys are pajamas to the worst extent in black sea.

From a game design perspective:

1. Group suppression needs to be replaced with individual suppression. This is made obvious when in "assault" during bounding overwatch. If the manuever elements gets suppressed the base of fire ellement will immediately start cowering. The whole job of the base of fire is to protect the manuever element and NOT lay down and give up. The guys manuevering are relying on the base.

2. Combat spacing should be maintained.

3. Reaction to contact drills would be nice. Taking fire from the front? Get out of your stupid file and get ON LINE and return FIRE

CMSF is more like this; US units (and Syrian paratroopers) are resilient under fire, for the most part.

This changed with CMBN for me. I was surprised to see platoons getting suppressed and then taking significant casualties from fire a few hundred metres away.

The only thing I would disagree with it is not just US troops. I'm doing the Russian campaign and it's making me very tentative in my attacks as infantry cower even under modest fire when they dash. A few times I've had to rescue them with the supporting APC, which they are meant to be screening!

I feel the small arms modelling could be dialled down a little bit.

Edited February 14, 2015 by Sulman

[Combatintman]

This one Carl ?

http://community.battlefront.com/topic/118052-prioritizing-the-scenario-editor/?p=1580979

p

Yes that would be the one ... although even I'm too young for OGs!!!!

[dan/california]

CMSF is more like this; US units (and Syrian paratroopers) are resilient under fire, for the most part.

This changed with CMBN for me. I was surprised to see platoons getting suppressed and then taking significant casualties from fire a few hundred metres away.

The only thing I would disagree with it is not just US troops. I'm doing the Russian campaign and it's making me very tentative in my attacks as infantry cower even under modest fire when they dash. A few times I've had to rescue them with the supporting APC, which they are meant to be screening!

I feel the small arms modelling could be dialled down a little bit.

Do you have any idea how many hundreds of discussion THREADS, many multi page ones, there were about turning it up?!

 

Post after, post about insufficient suppression from machine guns, and on, and on.  Glad to see nothing has changed, the board is never happy.

[Melchior]

People are just never happy that the game doesn't guarantee them a result they think they were entitled to. 

[sburke]

Edit: one frustrating things though about this is how the Battlefront fan boys can't admit that this game shouldn't have these problems to begin with, hence they call you a crap player and then tell you to micro the **** out of your troops as a work around to the problem (that isn't supposedly there to begin with).

I gotta ask, you've been a forum member for 3 weeks.  How do you jump to a conclusion like that in 3 weeks?  Personally I consider myself a "Battlefront fan boy"  I love the game, I admire the company and generally I think most things in the game work well more often than not.  I also feel most of the complaints on the forum come from not being able to work within the constraints of the game using the tools already included.  That being said, yes there is always room for improvement and there are some things that just don't work very well within the constraints of the game.  So does that make me a negative blight on the forum? 

 

I do believe that the nature of the game does make it that "micro managing" your troops makes things work better.  If it didn't I could just order my entire force with one command to attack an objective and sit back while the tac ai did all the work (and be bored to friggin tears in the process).   There is a reason you can split teams into different structures depending on your current objective and it works pretty darn well besides being hugely satisfactory when you do it right.

[Pelican Pal]

He could have been lurking and just recently registered.

 

FWIW, I'll throw my hat in with Oakheart on that one. I've seen numerous occasions where the militia rolls out on folks.
 

 

I do believe that the nature of the game does make it that "micro managing" your troops makes things work better.  If it didn't I could just order my entire force with one command to attack an objective and sit back while the tac ai did all the work (and be bored to friggin tears in the process).   There is a reason you can split teams into different structures depending on your current objective and it works pretty darn well besides being hugely satisfactory when you do it right.

 

 

This is essentially a light version of what he is talking about.

 

 

Edit:

 

The infantry game in CM is my favorite part of the series. I personally don't enjoy vehicle combat all that much. But I will say that the infantry combat is pretty bad and needs the most work relative to everything in the game right now.

 

There are just loads of problems with it.

 

The tight formations which make automatic weapons absurdly effective, the lack of SOPs which causes the tacai to do stupid stuff like run at the guy who is shooting them, the lack of any decent assault order, etc...

 

You can get around a lot of these issues once you understand the failings and you can generally get things to work out in a positive manner. However, that doesn't mean that just because we can work around the problems the problem stops being a problem.

Edited February 15, 2015 by Pelican Pal

[Sulman]

People are just never happy that the game doesn't guarantee them a result they think they were entitled to.

I don't think that is fair, and it says nothing about the merit of the points people are making.

Personally I can live with these things and I really like Black Sea, but it's healthy to look at them.

Post after, post about insufficient suppression from machine guns, and on, and on.  Glad to see nothing has changed, the board is never happy.

I was busy playing Shock Force. I only knew that happened in CMBN when I read the patch notes. I actually think CMSF had infantry about right - if you didn't look after them they'd punish you for it. I do think the crewed MG changes specifically were good because they are pretty useless in CMSF.

[Melchior]

I don't think that is fair, and it says nothing about the merit of the points people are making.

 

They aren't making a point when all they're doing is whining.  

[AttorneyAtWar]

They aren't making a point when all they're doing is whining.  

 

I don't see anything that can be considered "whining" except in one or two posts. These are all valid criticisms, and they have been a problem that has been brought up many times on this forum over the years.

[sburke]

He could have been lurking and just recently registered.

 

FWIW, I'll throw my hat in with Oakheart on that one. I've seen numerous occasions where the militia rolls out on folks.

 

 

This is essentially a light version of what he is talking about.

 

 

Edit:

 

The infantry game in CM is my favorite part of the series. I personally don't enjoy vehicle combat all that much. But I will say that the infantry combat is pretty bad and needs the most work relative to everything in the game right now.

 

There are just loads of problems with it.

 

The tight formations which make automatic weapons absurdly effective, the lack of SOPs which causes the tacai to do stupid stuff like run at the guy who is shooting them, the lack of any decent assault order, etc...

 

You can get around a lot of these issues once you understand the failings and you can generally get things to work out in a positive manner. However, that doesn't mean that just because we can work around the problems the problem stops being a problem.

 

So disagreeing and citing how the game mechanics work immediately qualifies one for being dismissed as a fanboy?  Hell in that case going forward maybe we'll all just be silent when you ask for advice as it seems the only option is to hop on your (I mean that in a generic sense, not you specifically) bandwagon and agree.

 

I too prefer the infantry aspect more, heck just look at my AARs.  I would however seriously take issue with what you consider as "broken".  The game has constraints, they are not based on real world dynamics, they can only approximate them.  Within the game constraints that approximation is pretty darn good.  Would I like more, hell yes.  I'd love to be able to peek around a corner before sending a whole team into an ambush. I'd love a bazooka team be able to lean around a corner and fire. That I can't doesn't mean the game is broken, it just means I can't. 

 

This is more the point, understanding that what you want in a game for coding reasons isn't there does not translate to broken.  The supposed lack of a decent assault order is another good one.  You want the TAC AI to do a very complicated maneuver where you just click assault and objective and the game does it for you.  That it does not and instead you have to figure out how to split teams and issue commands to get the job done does not make it broken. It means you have to learn what is possible in the game.  That you don't want to is your choice, but the option has been provided.

 

When you guys figure out how to program a better TAC AI, perhaps I'll sit up and take notice and stop being such a fanboy, but as far as I know, the TAC AI in CM is pretty much the best thing out there in any comparable game.  Will we keep asking for more, absolutely.  Does Steve want to do more, absolutely. Is the TAC AI a veritable sinkhole for resources and change for it going to come pretty slow, absolutely. Learning patience and finding alternatives as the game develops is not the end of the world.  Being asked to have that patience and understand what might be possible given the current state of the engine does not entitle one to start flinging what you consider to be derogatory names. 

[Pelican Pal]

So disagreeing and citing how the game mechanics work immediately qualifies one for being dismissed as a fanboy?  Hell in that case going forward maybe we'll all just be silent when you ask for advice as it seems the only option is to hop on your (I mean that in a generic sense, not you specifically) bandwagon and agree.

 

Haha, I was just saying that reply is a very light example of the "your complaint is wrong. The game is fine. You are just playing badly" thing he was talking about. I mean you essentially just misrepresented his position and claimed he wants the AI to handle everything for him as a way of proving that how things currently work is fine and dandy. Now I haven't read most of his posts, but I think he just wants the tacai to handle things that historically BFC wants the tacai to handle.

 

 

I too prefer the infantry aspect more, heck just look at my AARs.  I would however seriously take issue with what you consider as "broken".  The game has constraints, they are not based on real world dynamics, they can only approximate them.  Within the game constraints that approximation is pretty darn good.  Would I like more, hell yes.  I'd love to be able to peek around a corner before sending a whole team into an ambush. I'd love a bazooka team be able to lean around a corner and fire. That I can't doesn't mean the game is broken, it just means I can't.

 

I don't think anyone is saying the game is broken. Certain things are definitely a weakness. Infantry combat is one of them.

 

 

This is more the point, understanding that what you want in a game for coding reasons isn't there does not translate to broken. 

 

I didn't say the game was broken. Seriously, I'm still buying the damn things.

 

 

The supposed lack of a decent assault order is another good one.  You want the TAC AI to do a very complicated maneuver where you just click assault and objective and the game does it for you.  That it does not and instead you have to figure out how to split teams and issue commands to get the job done does not make it broken. It means you have to learn what is possible in the game.  That you don't want to is your choice, but the option has been provided.

 

No, I do not want the AI to do a complicated maneuver for me. I just want an order that prioritizes shooting someone in the face over running past them. I can't tell you the number of times my men have been moving and seen an enemy within 10-20 meters of them. They run past him as this enemy soldier does that glacial turn, picks a target, raises his rifle, and shoots someone. At any point during this 2-3 seconds they could have stopped and shot him in the face, but they don't cause they're too busy running.

 

Close combat is a definite weakness in the game and while we have a plethora of orders that handle movement/combat reactions we don't really have any geared for CC. An order that says "hey, you need to walk over there, but!, there might be a bad dude with a gun. So if you see him, you should shoot him immediately." would be nice.

 

Now are there ways around this deficiency? Yea, you move a bit slower, use teams, recon by fire, overwatch, etc.. So I do that and it works pretty well. Doesn't mean I wouldn't really love an assault order.

 

I don't think patience means not saying that this stuff needs work. I mean I, and presumably all the folks posting on this part of the board, are still buying the bloody game. If I didn't have patience I wouldn't have thrown BFC $55 a week ago for CM:BS. And I'm not expecting to see any of this fixed in the near future, I'm a computer science major, I understand that this stuff is hard because I've done it before (not in the literal sense of creating a complicated AI that handles combat situation, but I have written AIs and its not an easy thing to do). Frankly I would feel absolutely blessed to see any improvements by the end of 2016. Personally I doubt we'll see anything for longer than that. Although an SOP system would probably be relatively easy so maybe we'll see that by the end of 2016.

Edited February 15, 2015 by Pelican Pal

[Wiggum15]

What do you think, do automatic weapons need to be nerfed a bit ?

Should lonely or isolated soldiers route faster ?

I would say yes.

The Formations (the "Death train") are much much tighter as in real life, also, soldiers always try to get to a specific part of that formation which results in unnecessary movement.

Then you have the "isolated shooter kills my whole squad" cases and you wander why is this guy still there, why does he still fight ?

Is he of fanatic motivation ?

[womble]

I've seen so many pTruppen killed by not going to ground fast enough when they get "near" their objective. Say it's a team hopping over a low wall to get away from suddenly-appearing fire from the side they're on. If they just got up, jumped the wall, hit the dirt and crawled to their favourite square couple metres in the destination AS, a good percentage of them would still be alive. As it is, they're over the wall before aimed fire can drop them, but they stutter about, looking for a place to lie down and the enemy saves them the indignity of landing on a squad mate by making them dead when they do so. The TacAI obviously knows it's moving to escape incoming, and it's made the right decision in getting to the other side of the wall. But as soon as it's in the destination AS, it seems to forget the purpose of its move.

[c3k]

What do you think, do automatic weapons need to be nerfed a bit ?

Should lonely or isolated soldiers route faster ?

I would say yes.

The Formations (the "Death train") are much much tighter as in real life, also, soldiers always try to get to a specific part of that formation which results in unnecessary movement.

Then you have the "isolated shooter kills my whole squad" cases and you wander why is this guy still there, why does he still fight ?

Is he of fanatic motivation ?

 

NOT directed at Wiggum15, but upstream a bit the term "fan boy" is being thrown around. It can be taken as an insult. As well, any refutation of a criticism can seem like an insult.

 

Is the game perfect? Not even close.

 

Is there a better tactical game at this scale? Not that I know of.

 

Example: I am playtesting a battle on 16 square kilometers. I have about 2 battalions against a similar force. (CMBN) One of my halftracks, filled with men, is advancing over hilly, wooded terrain. A 'schreck takes out the halftrack from close in. The surviving infantry piles out. The 'schreck team opens up on them, killing one. The rest, shocked by the impact, the bail out, and the ensuing MP40 burst, decide to hit the dirt, fire a few rounds, then run off down the backside of the hill. (In my imagination, the 2 man crew of the halftrack took the full impact of the 'schreck and are still sitting there in the cab, charred, as the halftrack burns.) THIS ALL HAPPENED WITHIN A SINGLE TURN. It was perfectly modelled. Perfectly. This was one little micro-vignette within a 4 battalion, 16 sqkm, 4 hour battle. Oh, over the next 3 turns, my squad recovered their morale, and I split it into 3 teams. After establishing some area targeting suppressing fire, I maneuvered the other two teams to either flank, and killed the 'schreck team. Then I applied buddy aid to the dead squad member. Micro management? Hell yeah. Satisfying that it worked perfectly? Even more so.

 

The TacAI and programming was phenomenal for the ambush and reaction. Again, it was perfect.

 

 

Now, onto the point of this thread...yes, it can be improved. There are situations when the actions aren't perfect.

 

To Wiggum's post:

 

For those who are new, there were long-time cries that automatic weapons were unrealistically ineffective. Now, there are complaints that they are too effective. A perfect balance is impossible. Really. What works in one case would be wrong for another.

 

Individual soldiers, out of command, surrender VERY easily. They are often ineffective with any incoming fire. It's easy to capture them. I don't see an issue with them needing to be nerfed any more. (Assuming "individual"="isolated" or "alone".)

 

Formations ARE too tight. Each team has to be able to locate inside one action spot. That is a game-engine limitation. Changing that would be a tremendous effort with HUGE ramifications to every facet of the game. (As it's been told to me.) A workaround is to break up your squads when the terrain gets complex. As well, do NOT set long movement orders. That strings them out. Waypoints let them accordion together.

 

Isolated shooters are sometimes fanatic. You don't know when, so that you can't take advantage of them. "Oh, it seems Sergei is now a Fanatic. Excellent. I'll use him in the tower."

 

 

Are the above items workarounds? Yep. There are shortcomings, but they are not gamebreakers. Every tweak has ramifications. In some cases the TacAI response seems wrong. In others, it could not be better.

 

You get out what you put in. To me, the benefit is down low, at the individual level. I've got to command at that level to get the most out of the game. See my halftrack tale. 

 

 

 

Ken

[alwaysfish]

One of the best things Battlefront could do is to allow players script AI. No game is perfect and no game will fit everyone's needs. And i think game company should not even dedicate resources to make game perfect. But to make everyone happy, developers could create a simple scripting language, where players would be able to script more complex, "high-level", commands using EXISTING "low-level" commands. That way micro-management could be reduced to just one command. Furthermore, developer, instead of spending time on implementing various military tactics because their loyal players demand that, could just focus to extending low-level commands, while community will be able to experiment with high-level commands. Win-win situation. There are so many successful games which took the approach of letting community "fix" imperfections of the game through scripting and modding. I believe that Battlefront has one of the most dedicated communities, BUT limited development resources. Solution - allow COMMAND SCRIPTING.

 

Very simple example: Player selects community developed command "Charge, charge!!!" and targets it to the building. Script divides squad into Assault and B team, orders B team to overwatch target building, orders Assault team Quick run to point of entry to the building, orders Assault team to wait 15 secs, orders Assault team to Quick run into building, orders B team to Quick run into building, orders Assault and B team to merge back into squad. 

[ASL Veteran]

One of the best things Battlefront could do is to allow players script AI. No game is perfect and no game will fit everyone's needs. And i think game company should not even dedicate resources to make game perfect. But to make everyone happy, developers could create a simple scripting language, where players would be able to script more complex, "high-level", commands using EXISTING "low-level" commands. That way micro-management could be reduced to just one command. Furthermore, developer, instead of spending time on implementing various military tactics because their loyal players demand that, could just focus to extending low-level commands, while community will be able to experiment with high-level commands. Win-win situation. There are so many successful games which took the approach of letting community "fix" imperfections of the game through scripting and modding. I believe that Battlefront has one of the most dedicated communities, BUT limited development resources. Solution - allow COMMAND SCRIPTING.

 

Very simple example: Player selects community developed command "Charge, charge!!!" and targets it to the building. Script divides squad into Assault and B team, orders B team to overwatch target building, orders Assault team Quick run to point of entry to the building, orders Assault team to wait 15 secs, orders Assault team to Quick run into building, orders B team to Quick run into building, orders Assault and B team to merge back into squad. 

Post number 2 eh?  Welcome to the forum.  Just so you know, even the actual game AI barely has any sort of command scripting as you describe let alone giving that power to the player.  I would be very happy if they eventually gave the AI that level of command level scripting, but know that what you ask for would probably consume the majority of BFC's meager programming capacity for more than a year or two.