Panzerschreck/Bazooka Range and AI Behavior.

[TheBlackHand]

Still learning the game, playing the training missions. Taking it slow.

Last night I had a schreck team open up on my M8 from 250+ yards away. Of course, all of his rounds were falling short and the team never moved. I let my M8 sit there and shoot back while the schreck wasted rounds. I thought the team would bug out, but they stuck around and stayed alive . . . until the Shermans showed up and finally put the panzerschreck to bed.

Question: why is the AI set to open fire on targets which are well beyond the maximum effective range? Also, why would the AI not be programmed to move it's AT team after the first shot or two and when they KNOW they have been spotted (and are being fired on by both a fifty cal and a cannon)? It seems that self-preservation has not been included in their thought process. Isn't the AI supposed to mimmick human behavior? I can't imagine that any human being would play the way the AI played in this situation. Not a seasoned player, anyhow.

I am not a programmer so I am not questioning the programming ability of the Battlefront people. It just seems to me (an layman/idiot) that if in the CMx1 games, there was a calculated "chance to hit/kill", it could be written in that no AI/AT weapons would open fire on a target unless there was a greater than, say 60% chance of a hit/kill in CM:BN. I suppose the old "target lines" argument may come in to play here. Therefore, I'm not asking that the old CMx1 info to be displayed (I'm assuming it isn't, I have not played against tanks yet). I'm just wondering if it could be possible to make this information available to the AI . . . and if they can have it, why couldn't we? If I were a tank commander/AT gunner I would certainly calculate my "percentage to hit/kill" on any target that I was considering firing on. It would also seem logical to program the teams to move after the first couple of shots, instead of hanging around to get pummeled by the armor that they are wasting their ammo on.

This type of programming would also solve the rare, but incredibly aggravating situation where the AI makes a ridiculously lucky shot from well beyond max effective range.

Again, I would assume that any human player would know better than to make these AI mistakes, which is why playing a human being is when the real battle actually begins . . . but couldn't the AI be improved in this way as well?

It just seems like common sense to me.

Also, the only other thing that bugged me about the battle I played last night was that the first HQ team that I sent up to spot, would not move to the front of the building and therefore did not spot the MG teams, AT guns, etc. I spent three turns trying to get them to move to the proper windows, but they always stayed at the back of the room, looking out the side windows. When the reinforcements showed up, I ordered the other HQ team up there, and they went right to the front windows. What happened here?

It seems like the movment options where much more precise in CMx1. I know the fanboys will jump down my neck and tell me how wrong I am and that this is impossible . . . but perception is reality. I don't remember this problem in that game. Hoping to hear that it isn't going to happen very often in CM:BN. I'm open to the notion that I need to get used to how to properly use the CM:BN engine and that is what I am doing.

[YankeeDog]

Yes; this has been brought up several times already. In addition to Shrecks and Zooks opening up at ranges where they have very little chance of hitting, SMGs also have a habit of blowing all their ammo at ranges of 200m or more, where they have limited effect. These two may actually be related, since the both reflect soldiers shooting at ranges where they can technically hit and kill the target, but the % chance of a hit is very low.

The SMG issue BFC has acknowledged and said will be fixed soon in a patch. I don't recall anything official from BFC regarding zooks and shrecks, but I'm cautiously optimistic they've noticed the comments on this, and this behavior will be adjusted in an upcoming patch.

[TheBlackHand]

Hope so. That's kinda why I'm adding my two cents. This seems like a fixable issue. It would improve the experience dramatically. I would expect the AI to behave this way on "Boot Camp" setting, because it makes it so easy for the human player to pinpoint and destroy them. Maybe the difficulty slider could apply to AI behavior as well. Tonight I will begin playing at the higher level of difficulty (probably "Warrior" setting, as it sounds like the most enjoyable setting to me).

[Blackcat]

"first HQ team that I sent up to spot, would not move to the front of the building and therefore did not spot the MG teams, AT guns, etc. I spent three turns trying to get them to move to the proper windows, but they always stayed at the back of the room, looking out the side windows. "

Sorry to ask the obvious, but did you use the face command? I find telling a unit in a building to face the way I want them to look works.

P.S. There is no difficulty slider in CMx2 just variations on the level of FOW.

[womble]

It seems like the movment options where much more precise in CMx1. I know the fanboys will jump down my neck and tell me how wrong I am and that this is impossible . . . but perception is reality. I don't remember this problem in that game. Hoping to hear that it isn't going to happen very often in CM:BN. I'm open to the notion that I need to get used to how to properly use the CM:BN engine and that is what I am doing.

Without wanting to jump down your neck...

The movement options in x1 weren't more precise, objectively, since the movement grid was based, I believe, on 20m squares, rather than the 8m squares that x2 uses. 2.5 times more precise! The problem didn't happen in x1 because of the imprecision/abstraction: the unit was assumed to be in the middle of the tile and got its LOS/LOF from there. In x2, you have the individual placement of troopers within the tile, and your control over those is necessarily less precise than you could at some times desire, in order to allow the AI to seek good micro-cover. But the combination of move and Face commands will usually allow you to get your truppen looking out the windows you want them to.

The reason the second HQ went to the 'right' set of windows is that the sub-positions for the 'wrong' set of windows were already taken. If you'd successfully used a Face command to get the first element looking the right way, the second element would have been forced to look out the other windows.

[Broken]

Without wanting to jump down your neck...

The movement options in x1 weren't more precise, objectively, since the movement grid was based, I believe, on 20m squares, rather than the 8m squares that x2 uses. 2.5 times more precise! The problem didn't happen in x1 because of the imprecision/abstraction: the unit was assumed to be in the middle of the tile and got its LOS/LOF from there.

Not true. The terrain may have had 20m resolution in CM1, but LOS and waypoint resolution was much finer than that. Shifting a unit by 2m could move you from in-LOS to out. That was possible because units were abstracted as being located at a single point, not spread over 8m x 8m as in CM2.

[MikeyD]

Shreks do indeed have a 300m range and zooks have a 250m range. So the question is should a team attempt to use its weapon at the outer limits of it capability? I can imagine an alternative complaint if it was restricted. "My shrek has a 300m range. An M8 is sitting in the open 250m away. But my team refuses to fire. What gives?"

I believe rocket launcher and SMG useage at longer range may have been tweaked some for the coming patch but I've got no detail on the changes.

[YankeeDog]

It's really apples to oranges when you're looking at the way CMx2 handles infantry unit movement and position, vs. CMx1.

Infantry units in CMx1 were singularity "electron clouds", a point entity abstractly representing a group of soldiers roughly centered around the plot point. From a technical standpoint, individual soldiers existed only in the sense that infantry teams lost firepower by quantum increments, each quanta representing the loss of an individual soldier.

Infantry units in CMx2 are composed of soldiers specifically and individually modeled, and the player picks an action spot or spots for the group to deploy upon. It's a far higher fidelity modeling system.

[TheBlackHand]

Sorry to ask the obvious, but did you use the face command?

Yes, I did. It didn't seem to work at all.

The reason the second HQ went to the 'right' set of windows is that the sub-positions for the 'wrong' set of windows were already taken.

Aha! That figures. I did, indeed have the first team up there as the second team placed themselves.

In CMx1 I could've placed the (moveable) waypoint at the corner of the building/second floor and thats where they would've gone. I do appreciate the answers giving me the technical reasons why units in CM:BN behave differently. Honestly though, it all begins to sound like algebra to me . . . and I'm no good at math. It goes back to what I said about "perception is reality". My perception is that, in certain respects, the pinpoint placement of units CMx1 was more precise. I understand that this may not actually be true.

In that regard, is there anything that will give me an indication of the amount/type of cover that my soldiers are currently utilizing? I haven't noticed that in the interface. Will look for it tonight. Would be a nice thing to have.

Shreks do indeed have a 300m range and zooks have a 250m range. So the question is should a team attempt to use its weapon at the outer limits of it capability?

Personally, I would say "no". This would be a ridiculous waste of ammunition and it would give away your position. That is the command that I would assume would most normally be given by any human commander. Therefore, that is the restriction that I would have coded into the game. Alternately, I would set it up so that if the PLAYER/COMMANDER gave a DIRECT order to that team (via cover arc, for example), then they could/would open up.

I do not think it is realistic or advantageous for AI teams to be using a weapons maximum range as their cue to open fire. I wouldn't think that any target sitting at max range would provide a better than 60% chance of hit/kill.

[herr_oberst]

Probabilities aside, I have had a 'shrek team land a top hit on a Sherman from 230 meters on the third try while playing 'Barkmann's Corner', a great little scenario. For real fun, imagine yourself as ONLY Barkmann. You can't touch or change the infantry placement or actions... it's just you and your Panther.

Try that one for fun, fun, fun.

[Broken]

Probabilities aside, I have had a 'shrek team land a top hit on a Sherman from 230 meters on the third try while playing 'Barkmann's Corner', a great little scenario. For real fun, imagine yourself as ONLY Barkmann. You can't touch or change the infantry placement or actions... it's just you and your Panther.

Try that one for fun, fun, fun.

A 230 meter kill is amazing. My best was a 160m first-shot kill of a Panther with a M9A1 bazooka. The Panther had just nailed a Sherman through smoke 20 seconds earlier.

[Jarink]

I would think unit experience should have a major effect on the range at which a Shreck/Zook crew opens fire. I would think lower quality troops would be more likely to open fire from too far away, as they would not have the 'nerve' or practical experience to know to hold fire until the target is at a range that gives a good chance of a hit. There's plenty of historical examples of this behavior over the past couple hundred years.

[panzermartin]

Amazing I just got a ~ 290m first shot- kill with a shreck! Sometimes its good to just let it fire

[Panzerfest]

Shrecks are great because they have a long reach, Ive been plugging tanks 200m + routinely!

[TheBlackHand]

I would think unit experience should have a major effect on the range at which a Shreck/Zook crew opens fire. I would think lower quality troops would be more likely to open fire from too far away, as they would not have the 'nerve' or practical experience to know to hold fire until the target is at a range that gives a good chance of a hit. There's plenty of historical examples of this behavior over the past couple hundred years.

I agree with this, but I still think that even an inexperienced crew would know better than to open up at distances such as . . . .

Ive been plugging tanks 200m + routinely!

If true, this should not be happening. Certainly not "routinely".

Seems a simple fix, relatively speaking.

[gunnergoz]

Regularly hitting anything over 100-150M away - even a tank - with these early rocket launchers sounds excessively accurate to me. I'm doing some research into it to see what the CEP on them was but data is sparse.

[Broken]

Regularly hitting anything over 100-150M away - even a tank - with these early rocket launchers sounds excessively accurate to me. I'm doing some research into it to see what the CEP on them was but data is sparse.

Even the more accurate bazooka, M9A1, is best used under 80 meters for a high hit probability from a Regular un-rattled crew. Hitting at 200 meters is probably <5%. Hitting at 250m is winning the lotto.

[dieseltaylor]

In a letter dated May 20, 1944, Gen. George S. Patton stated to a colleague that "the purpose of the bazooka is not to hunt tanks offensively, but to be used as a last resort in keeping tanks from overrunning infantry. To insure this, the range should be held to around 30 yards."[23]

Ha! shows what he knew.!

[Doug Williams]

I do not think it is realistic or advantageous for AI teams to be using a weapons maximum range as their cue to open fire.

I agree. While there is a small chance that the zook or shreck will hit the target at maximum range, 90% of the time what will happen is the AT team will simply give away it's position too early, and then become the object of attention of a 30-50 ton hunk of steel sporting machine guns and cannons that will soon turn the AT team into a fine red mist.

[Nidan1]

Shreks do indeed have a 300m range and zooks have a 250m range. So the question is should a team attempt to use its weapon at the outer limits of it capability? I can imagine an alternative complaint if it was restricted. "My shrek has a 300m range. An M8 is sitting in the open 250m away. But my team refuses to fire. What gives?"

I believe rocket launcher and SMG useage at longer range may have been tweaked some for the coming patch but I've got no detail on the changes.

Are we talking about the AI or are you suggesting that a human controlled team would not fire at longer ranges, if the software was altered?

[Blackcat]

Are we talking about the AI or are you suggesting that a human controlled team would not fire at longer ranges, if the software was altered?

There is but one TacAI in the game. It is used both by the computer and human player. If the logic is changed on when a unit will open fire it will apply to both sides. The human player can, mostly, over-ride the TacAI's decisions on targetting, assuming he knows that there is a decision to take and is in a position to take it - not always guranteed in real time or WEGO.

The call for a hard limit on when the computer player will open fire with an infantry AT weapon is, I think, fraught with difficulty. For example, I know that the effective range of such a weapon is generally less than 100 yards, but it could kill me at more than twice that range. Now, if there is a hard limit of, say, 100 yards then I can place my tank at 101 yards distance and sit there blowing the snot out of anything I fancy in perfect safety. Getting a good LOS at 250 yards is, in Normandy, much harder than a LOS at 101 yards. So what is intended to be a help for the computer player could end up with being a major hinderance. The logic needs to be sufficiently fuzzy and that will always produce arguments on specific cases.

I don't say that the OP's idea is wrong just that it is not as simple as some appear to be thinking.

[YankeeDog]

IIRC, the Germans considered the Panzershreck's *effective* maximum combat range against a stationary tank to be about 150m, and against a moving target somewhat less. And soldiers were specifically trained to hold fire with the weapon until this range or less. IIRC, the American expectations and training regarding the effective range of the Bazooka was proportionally similar in relation to the weapon's theoretical maximum range.

So it seems to make sense to me to limit the TacAI's max. range for opening up with these weapons of its own accord to roughly the ranges that were actually historically taught to soldiers. Player could always override these setting with Cover Arcs and/or direct Target orders.

It would also make sense to me that inexperienced and/or poorly trained (i.e., Green or Conscript) and/or rattled units might open fire prematurely due to nerves, lack of training, and/or a poor estimation of distance. If this can be coded in, so much the better.

IMHO, the above TacAI protocols would make rocket IAT teams easier to manage for the player, and make the computer player a bit more of a challenge.

Cheers,

YD

[dieseltaylor]

A hard limit need not be a simplistic range limit. It might take into account% hit chance, and a disincentive for shooting head on.

When to shoot:

Therefore side on tank target 50 yds- x yards +20% to base level

Rear on +30%

Head on - 25%

Moving fast -10%

Stationary +20%

Immobilised +30%

Supporting troops nearby +20%

And likelihood to shoot tails off with range - along with hit chance

[Blackcat]

A hard limit need not be a simplistic range limit. It might take into account% hit chance, and a disincentive for shooting head on.

When to shoot:

Therefore side on tank target 50 yds- x yards +20% to base level

Rear on +30%

Head on - 25%

Moving fast -10%

Stationary +20%

Immobilised +30%

Supporting troops nearby +20%

And likelihood to shoot tails off with range - along with hit chance

Yup it could and it does need something like that. However, I thought that CMBN doesn't calculate % to hit chances! So lots more work to bring in what seemed to be a simple request.

[womble]

Yup it could and it does need something like that. However, I thought that CMBN doesn't calculate % to hit chances! So lots more work to bring in what seemed to be a simple request.

It doesn't, but I think it's fair enough to use 'hit chance' as shorthand for whatever algorithm the AI uses to figure out how likely it is that something succeeds.