Stupid STUG AI behavior

[Marc Anton]

Can we please rethink the AI behavior for STUG crews when using the secondary MG on the roof ?

These idiots always pop their heads out even when the STUG itself has the "button" command.

They will do that (and die of course) until 2 crew members are left ... and starting to panic .... ridiculous.

Its a very absurd behavior cause the player cant do anything else than not sending them into action .. but its a tank and it needs to fight.

[Oddball_E8]

armor only arc?

[mvp7]

Same thing with some armored vehicles and halftracks even at long ranges. Would be nice if they were a bit more careful when there's two or three dead would-be gunners in the vehicle.

[waclaw]

Somebody explain to me why so late to see them?

[sburke]

don't know for sure, but one thing it would be good to know - I can hear the crickets so it is apparently at best dawn and maybe earlier. What is the overall LOS? There is really little to work with from the video and it is not narrated in English so I can't tell if you are supplying additional info or not.

[LUCASWILLEN05]

Might level of traiing be a factor there. I have never seen armoured vehicle crews pop their heads out of hatches when under fire, certainly not of their own volition. If I tell them to ope up they will take a quick look and duck back down again.

[eniced73]

Somebody explain to me why so late to see them?

From where the Stug is sitting does it have LOS the whole way across that field? If not then those armored cars would be coming from a blind spot lower down on the other side. Mix in the spotting cycles and I can see how that happened. If he did have LOS across that field, I have no explanation as they should have definitely been seen.

[MikeyD]

Dawn, hazy, the intervening hill crest and the X-tall grass make for something of a 'perfect storm' spotting-wise. Plus the StuG's low mounted gun. Just because the commander can see something doesn't mean the gun's LOF will clear the crest. Sometimes StuG fires its roof MG because that's the only weapon with LOF.

[poesel]

I guess what's happening here is the usual mixture of spotting intervals and bad luck.

Spotting is calculated every n (with n mostly 7) seconds. So every 7 seconds you have a chance to spot something. If you miss one chance it could well be 14 seconds of total blindness. That is enough time to travel quite a distance.

Reducing the spotting interval is a performance hit and chance is necessary to simulate real world events.

Having said that I would still argue for doing away with chance and intervals in very limited circumstances: when vehicles are very close.

It is simply impossible to miss a tank that's in plain sight 8m away from you (infantry spotter, not hiding, not suppressed). Still it (very seldom) happens and then it breaks immersion hard and you get those posts here.

I don't know how visibility was in the video above but the AFV and the tank were on the same path so the AFV should have been spotted in the position where the tank got nailed. But the Stug missed the AFV with his first chance and then spotted both on the the next interval simultaneously.

The spotting system works very well but not for these corner cases.

[Marc Anton]

armor only arc?

Not an option if there are NO enemy tanks around ....

But that is not the point.

Buttoned should mean buttoned, isnt that hard to understand that command !

[c3k]

I saw it happen to my guy, he died, and I liked it.

The StuG has NO close defense mechanism. If infantry approaches, the crew MUST get them away, or flee. Otherwise, the vehicle will be destroyed. Better to pop up, spray the troops, then button back up.

Of course, I did not look at the vid: if they were shooting distant targets, then ignore my writing, above.

In the one case I had, the enemy infantry were pretty close.

Ken

[Oddball_E8]

Not an option if there are NO enemy tanks around ....

But that is not the point.

Buttoned should mean buttoned, isnt that hard to understand that command !

Buttoned doesn't mean "never poke your head out".

It just means stay inside as much as possible, but pop out to fire the gun (on vehicles that have exterior guns).

You'll see the same behaviour on all vehicles with exterior guns.

[Oddball_E8]

As for the video, the commander might have been looking through binocolars at the time and the gunner/loader popping his head up looks like he has a pretty poor view of that area that the vehicle is coming from.

[LukeFF]

Buttoned doesn't mean "never poke your head out".

It just means stay inside as much as possible, but pop out to fire the gun (on vehicles that have exterior guns).

You'll see the same behaviour on all vehicles with exterior guns.

Buttoned does mean buttoned, or otherwise we'd be seeing Sherman TCs popping up to take shots with their .50 cal HMG when they've been given a Button order.

[Oddball_E8]

Buttoned does mean buttoned, or otherwise we'd be seeing Sherman TCs popping up to take shots with their .50 cal HMG when they've been given a Button order.

Not quite, since they do have a coaxial MG and a bow MG.

The stug has neither.

[Battlefront.com]

To me the vid shows a bunch of different factors combining into an undesirable outcome.

1. The T-70 is small, came from a fold in the terrain, and is moving VERY fast. This dramatically reduces reaction time from the StuG crew under any circumstance.

2. The player unbuttoned the StuG crew after the T-70 was on the move. This means the StuG now has an improved capability to spot, but the target is already up close and personal by the time the next spotting cycle comes about.

3. The now unbuttoned crew spots the T-70 because a) it's very close and the StuG now has maximum spotting potential.

Had the StuG been unbuttoned about 7 second earlier it is quite possible the T-70 would have been spotted earlier. But the short distances involved and the speed means it wouldn't have given the StuG much time to react in any case.

The #1 problem here is that AFVs, especially StuGs, are not designed to get into "knife fights". They are designed to engage targets at distances of hundreds of meters, not tens of meters. There are lots of good real world reasons for this, some of which are shown (perhaps a bit crudely) in this video. What I mean by that is even if the StuG saw the T-70 at the first possible second it possibly could (i.e. LOS possibility) I don't think the StuG would have been able to act any differently. We're talking about a few seconds at close range with a rapidly closing target that is moving at an oblique angle to a turretless vehicle.

My conclusion here is that what you see in the video might not "look" right, but the outcome is probably the same as if it were possible to have the game check spotting every millisecond. A StuG in this position in real life would be a massive disadvantage.

Unfortunately, no system is perfect 100% of the time all the time. Which means things like this will happen. Perhaps in the near future we will be able to syphon off some more CPU time to shorten the spotting cycles further. Until then, CM will continue to have the most detailed, most realistic spotting system of any wargame (and probably game) in existence. You guys are lucky that we've never been satisfied with being the best because otherwise we'd not bother making something like this better

As a note, ~7 second spotting time is shortened when units are in close proximity. I think it can go down to ~1 second IIRC. Unfortunately wherever the line is drawn there is a possibility for something to crop up that should be treated the opposite way that it is.

Steve

[poesel]

As a note, ~7 second spotting time is shortened when units are in close proximity. I think it can go down to ~1 second IIRC.

Good dynamic solution! Is that new with v3?

[A Canadian Cat]

Just in case Steve does not make it back here any time soon:

Good dynamic solution!

Agreed - very clever.

Is that new with v3?

Nope, Steve has mentioned this before (now I have no internal knowledge there could have been some changes I have no idea). If I recall correctly I think someone conducted some experiments and estimated that the spotting interval was 7s and Steve fill us in on the dynamic nature of their solution to help make things better when units are close to each other. That was a while ago too.

[Battlefront.com]

Yes, that's been in for a while now. A few years for sure. If you played the game prior to that you'd see instances of units (especially infantry) in tight terrain having targeting problems when in very close proximity to each other. And like the StuG video, units moving fast made things even worse. What Charles did was figure that there wouldn't be too much of a hit to the framerate to have more frequent close checks because close in combat isn't all that common and it is usually resolved quickly.

Steve

[Jock Tamson]

wouldn't be too much of a hit to the framerate

Personally I would rather have more frequent cycles at the expense of Real Time. I have a a very high end system and real time is not really much fun on anything bigger than a small battle. 15 FPS is typical for me in forests etc so Real Time is not a serious option.

[Battlefront.com]

Spotting cycles must be the same for everybody all the time for every battle in any mode. There's few things in the game that have such a huge across the board impact. We're not going to decrease the spotting cycle times until we have plenty of wiggle room to do it without degrading overall game capabilities.

Steve

[Amizaur]

As a note, ~7 second spotting time is shortened when units are in close proximity. I think it can go down to ~1 second IIRC. Unfortunately wherever the line is drawn there is a possibility for something to crop up that should be treated the opposite way that it is.

Steve

It would be also good to have spotting cycles shortened if a narrow cover-arc is defined.

I mean - more frequent spotting cycles for detecting enemies IN THE COVER-ARC.

It would increase chances of someone who is EXPECTING enemy to show up in some place, to spot first and shot first. The more frequent spotting cycles in the arc could be compensated by less-frequent spotting cycles outside of the arc, to make overal game-wide frequency of spotting cycles constant.

(Currently we can put a tank in an ambush - stationary, with narrow cover arc set where we exepct the enemy. But the enemy who shows up has exactly same chances to be the first who gets his spotting cycle and to spot - as the ambusher. So what's the point in making ambushes ?)

[akd]

That would not be good. That is not the purpose of target arcs.

[Bastables]

Yes, that's been in for a while now. A few years for sure. If you played the game prior to that you'd see instances of units (especially infantry) in tight terrain having targeting problems when in very close proximity to each other. And like the StuG video, units moving fast made things even worse. What Charles did was figure that there wouldn't be too much of a hit to the framerate to have more frequent close checks because close in combat isn't all that common and it is usually resolved quickly.

Steve

I was wondering why units were "reacting" better and i did not have that disquieting "bull**** not spotted but being killed anyway" moments in CMRT as with my forays in earlier CMx2 titles. Intuitive is not the right word but it's the first thing that comes to mind after only two CMRT battles.

Maybe the best changes for me personally especially with coupled with the hit decals.

[Battlefront.com]

A simulation is a tricky thing even when the subject matter is simple. The more you stress the system, the more likely it will yield an undesirable outcome. CM simulates an extremely complex subject matter and players have incentives to stress it out (i.e. "gamey" behavior). The more people try to engage in combat practices that are not realistic or common, the greater the chance that CM will toss out a result that doesn't seem believable. And since wargamers are Human (allegedly, at least ) they are prone to focusing on the times when such a result comes up disproportionate to frequency.

In other words, if a bad spotting event happens 1 in 10,000 times we will likely see a thread about it and a couple dozen responses. We will not see that for the other 9,999 time. And thank goodness for that

Steve