Jump to content

CMRT TacAI Engine Comparison


Recommended Posts

Having played with engine 4 for half a year now and getting to know how it works, I finally decided to make a video about the current behavior of the TacAI. The basic premise is that the new TacAI behavior is possibly bugged when it comes to their reaction to indirect fires. 

I'm starting a new thread, and in the CMRT part of the forum because the game I used to compare the differences in the TacAI is CMRT, and I am hoping that this thread will serve as a collection point for further discussion on the v4.0 behavior. 

 

A few notes on the video itself. I'm currently running two installations of CMRT. One with version 3 and one with version 4. My version 4 install is modded, whereas my version 3 is not. 

The scenario I am using is the second mission from the training mission in CMRT. I used the campaign unpacker tool to extract it, and added a single 4 tube battery of 82mm off map mortars to the Russian side. I also tweaked the veterancy, motivation, and leadership values of the Germans a few times to see if it yielded different results.

Each time, the fire mission I called in was the same; medium Rof and duration, area target that targeted the same spot every time. It was a first turn artillery barrage, so there was no need for spotting rounds. 

 

I understand that this video is not a conclusive test, I did not intend for it to be that. It is simply to show a comparison between common behaviors found in versions 3 and 4, and to show that many of the behaviors encountered in version 4 are not optimal. It's my personal opinion that there is a bug here, but again I have not run enough tests to accumulate the data necessary to say that for certain. 

If others have documented examples of the current 4.0 behavior, please feel free to post it here. With any luck, we can show that the behavior is at the very least a bit off, and it could help BFC in tweaking/fixing it. 

Link to comment
Share on other sites

  • Replies 64
  • Created
  • Last Reply

Top Posters In This Topic

This is a good test and illustrates perfectly the problems most of us are encountering. What is the point of entrenchments at all if troops abandon them so easily? In the other post on the FI forum Steve comments about the effects of panic on troops. I think we can all agree that if troops panic, no matter what their skill level they will become unpredictable. The problem is that troops are vacating buildings and entrenchments before they get to a panicked state as was aptly shown by the fleeing of the MG team in the video. I have seen this many times in my own battles. I think it needs more than just tweaking. Personally I would be for re-setting things back to version 3.0 at the very least. 

Link to comment
Share on other sites

35 minutes ago, Hilts said:

This is a good test and illustrates perfectly the problems most of us are encountering. What is the point of entrenchments at all if troops abandon them so easily? In the other post on the FI forum Steve comments about the effects of panic on troops. I think we can all agree that if troops panic, no matter what their skill level they will become unpredictable. The problem is that troops are vacating buildings and entrenchments before they get to a panicked state as was aptly shown by the fleeing of the MG team in the video. I have seen this many times in my own battles. I think it needs more than just tweaking. Personally I would be for re-setting things back to version 3.0 at the very least. 

Well, if you read what Steve is actually saying. You know that most of the troops thats leaving their cover, is not panicing. They are relocating. although some of them is in panic mode. But as you have seen in Steves post, he allso Thinks that this behavior needs a Little tweak. The few ones, that is actually panicing. That might be correctly, but the relocating? Thats what need a tweak., like Steve say. On an open field, if  there is a cover a Little bit a way. IRL thats where troops will seek cover. The problem now is that they leave the cover they already have, to run out in the field. Sometimes even against the enemy.

But it was very good Steve posted an answer. Now at least we know, that he allso Thinks. That some kind of tweaking is needed.

Link to comment
Share on other sites

I think the primary issue here is that units are deciding to break from cover and flee into the open when under indirect fire. I don't have a problem with infantry units making a break for a building or a foxhole if they get caught in the open. I do have an issue when they are already in a building/foxhole/trench, and decide to displace during the middle of a barrage. All it ends up doing is breaking up the cohesiveness of a position and more importantly, getting men needlessly killed. 

Hopefully the video I posted shows this behavior adequately. 

3 hours ago, Hilts said:

Personally I would be for re-setting things back to version 3.0 at the very least.

I have also come to the conclusion that the v3 behavior is better for right now. Its a shame because I really do like all of the improvements that came with the v4 upgrade.

Another point worth making on this matter: the pause command workaround (where you simply give all your men you do not want fleeing from their positions an indefinite pause order) doesn't always work. Not only is it very tedious to be managing everyones pause states, but what can happen is if a deployed MG team is given a pause command, a barrage lands near it and triggers the flight response but it does not flee due to the pause order, what happens is they will no longer man the gun. In order to get them back on the gun, you have to unpause them, and then hope that they do not flee. 

There is also some questionable behavior when it comes to fleeing from small arms fire as well. I may make a second video that illustrates these issues if I think there is a need.

Again, my goal is simply to showcase the behavior we are seeing, so that if it does need to be tweaked, the dev team has a better idea of what it is specifically that is happening that needs to be tweaked. 

Link to comment
Share on other sites

'Tree bursts' above troops are the most murderous. They're air burst detonations equivalent of having VT fused artillery dropped on top of you. Getting away from the tall trees is not an unreasonable plan of action. I'd be interested to see how this plays out on other terrain, among brush and rocks where the artillery isn't detonating 15 feet over your head.

Link to comment
Share on other sites

2 hours ago, MikeyD said:

'Tree bursts' above troops are the most murderous. They're air burst detonations equivalent of having VT fused artillery dropped on top of you. Getting away from the tall trees is not an unreasonable plan of action. I'd be interested to see how this plays out on other terrain, among brush and rocks where the artillery isn't detonating 15 feet over your head.

I've done simple tests in the past and troops will exit trenches placed in the open. It also seems this is something you see mostly with infantry occupying fortifications against off-map artillery.

Link to comment
Share on other sites

18 minutes ago, Xorg_Xalargsky said:

I've done simple tests in the past and troops will exit trenches placed in the open. It also seems this is something you see mostly with infantry occupying fortifications against off-map artillery.

+1

My experiences with this in the game weren't tests, but I've certainly had many instances where troops weren't running away because of tree bursts. Also, as I mentioned in my post back in January 9, there's an operational aspect to this behaviour. Foxholes may be readily abandoned, but trenches represent significant investment [historically, 'investment' was a military term that made its way into economics] by both the troops digging them and the command ordering them to do so, and should not be abandoned unless troops holding them outright panic. [And WW1 serves to show that troops will hold out in trenches under hellish bombardment.]

Link to comment
Share on other sites

4 hours ago, MikeyD said:

'Tree bursts' above troops are the most murderous. They're air burst detonations equivalent of having VT fused artillery dropped on top of you. Getting away from the tall trees is not an unreasonable plan of action. I'd be interested to see how this plays out on other terrain, among brush and rocks where the artillery isn't detonating 15 feet over your head.

Right, but one is much more vulnerable running around while tree bursts are going off than they would be if they just stay put in their trenches. Again, the instinct when under artillery fires is to "get small," as in finding the nearest bit of adequate cover and staying there until the shelling lifts. This means dashing into a foxhole/trench/house, not out of those. The latter being what is currently happening in V4. Plus, and correct me if I am wrong, but I thought one of the reasons (besides having the benefit of being hidden from non-LOS) that trenches were reworked from terrain features in SF to what they are now is because now they provide a cover bonus that includes some cover from airburst. There isn't any overhead cover on the trenches visually represented, but it is simulated to an extent. 

All of this is besides the point however. Regardless of whether or not the fortifications are under tree cover, or in the middle of the open, the behavior is the same. The infantry, regardless of skill, leadership or motivation, run away from good cover, into the open when under artillery fire. It is completely nonsensical. 

Another quick note on my video; some may notice the mission time. The mission I was using was the second mission in the training campaign. There are 30 minutes (30 turns) in the mission. Note that for all of the tests, this behavior occurs within the first minute of gameplay. This behavior is not the result of infantry 'cracking' under a prolonged barrage. This behavior occurs when the very first shell lands close enough to an infantry team/squad. To further the point, if the behavior was so common and so readily triggered in just the first minute of this battle (again regardless of leadership/motivation/skill) it follows that others are experiencing the same behavior. 

I will likely make a follow-up video after getting more feedback, and thinking up a few new ways to show off the current behavior. For example, I intend to show that this fleeing behavior can be triggered by small arms fire. While not as serious a problem as the artillery fleeing, in my opinion it makes infantry a bit too brittle. In the follow-up video I will be sure to include a scene showing infantry in foxholes/trenches that are not under trees being shelled, and their reactions. 

 

Again, all of this is not meant to be a "damning criticism" of the game. I'm simply trying to show what I believe to be a bug with the hopes that showing it will help to fix it.  

Edited by IICptMillerII
Link to comment
Share on other sites

8 hours ago, MikeyD said:

'Tree bursts' above troops are the most murderous. They're air burst detonations equivalent of having VT fused artillery dropped on top of you. Getting away from the tall trees is not an unreasonable plan of action. I'd be interested to see how this plays out on other terrain, among brush and rocks where the artillery isn't detonating 15 feet over your head.

Foxholes and trenches provide near absolute protection against ground bursts, but continue to provide very good protection against air bursts.  The difference is actually probably marginal, especially for heavier ordnance, since the precision needed to deliver air burst effects down into a trench / foxhole would be fairly close to what would count as a direct hit on the position with ground bursts.

 

There are two problems here:

1. Core CM problem unrelated to Engine 4: AI has no concept of proactive use of cover and only uses cover reactively.  Foxholes and trenches will not provide good cover for AI until AI has already taken significant casualties from indirect fire.

2. Engine 4 problem that exacerbates the above: AI in cover will proactively displace from cover after receiving very light casualties.

You can see the feedback loop problem that results.  Solution: instead of AI recognizing indirect / HE threat and proactively displacing, AI should recognize foxholes / trenches as good cover versus indirect / HE and proactively use this cover when subjected to indirect / HE (cower / hide to minimize exposure).  Artillery vs. AI in cover then becomes more about suppression than destruction (as is the case in reality).

Edited by akd
Link to comment
Share on other sites

9 hours ago, MikeyD said:

'Tree bursts' above troops are the most murderous. They're air burst detonations equivalent of having VT fused artillery dropped on top of you. Getting away from the tall trees is not an unreasonable plan of action.

Actually trees protect quite well against artillery in this game.

Because even though the trees correctly make bursts more dangerous by moving point of detonation up above ground level, either the trunks block many of the LOS checks to individual troops, or there's simply a protection bonus for being in wooded squares that more than makes up for the tree burst effect. At least with 60mm mortars and smaller bombs.

But I've also seen a 150mm shell burst in a small tree directly above one of my scout teams. One man was lightly wounded.

Link to comment
Share on other sites

4 hours ago, IanL said:

Very cool work, thanks for doing this. I look forward to your small arms test. My request would be to also include the scenario file and a saved turn with the orders you gave for the video recording set. Being able to see the turn execute and repeat it is helpful.

I have some FI saved turns from a PBEM battle that perfectly illustrate the currently absurd situation of good quality troops fleeing foxholes and trenches due to enemy small arms fire. Let me know where you want 'em sent.

Link to comment
Share on other sites

12 hours ago, MikeyD said:

'Tree bursts' above troops are the most murderous. They're air burst detonations equivalent of having VT fused artillery dropped on top of you. Getting away from the tall trees is not an unreasonable plan of action. I'd be interested to see how this plays out on other terrain, among brush and rocks where the artillery isn't detonating 15 feet over your head.

As others have already iterated, this happens regardless of overhead. Units entrenched in a field will bolt from an off-map spotting round (general, non airburst).

I've sent @sburke saved files showing this problem and can corroborate - in fact, Miller was my playtesting buddy in those saved files. He hasn't pulled this out of nowhere, I assure you. We've been mulling over how to articulate the issue without getting cornered by ifs and whats. I'm confident enough now in saying that this isn't, as is often the case, players interpreting combat differently from the devs. This is a problem; one we hope the Dev team will tackle in due time with their usual skill.

The 'pause' command has circumnavigated most of this problem and keeps the game interesting H2H, its quite unplayable now SP; and I'm not willing to accept any hand-wave explanations that this is somehow 'rational' and comparable action in reality.

Edited by Rinaldi
Link to comment
Share on other sites

13 hours ago, MikeyD said:

'Tree bursts' above troops are the most murderous. They're air burst detonations equivalent of having VT fused artillery dropped on top of you. Getting away from the tall trees is not an unreasonable plan of action. I'd be interested to see how this plays out on other terrain, among brush and rocks where the artillery isn't detonating 15 feet over your head.

The data that has been provided has nothing to do with tree bursts.  The behavior is the same regardless of surrounding terrain.  As AKD noted above the question is more about what should be the AI reaction.  This isn't a new topic, just one that got tweaked recently based on suggestions/requests to deal with other situations where the AI just sits there and let's you plink them with a mortar.  There are actually situations where you would want the AI to have this behavior.  They just aren't getting the same attention as the impact overall seems to be more negative than positive.

It also isn't entirely new.  In developing the scenario Frosty Welcome for the MG module I was frustrated by how easy it was to cause the Brits holding the schoolhouse to just bail out into the street. (in reality was a heavily fortified position that held out for quite a  while).  That well predates 4.0.  I think the tweaks just made it more apparent.

I suspect there isn't an easy answer for the AI in single player mode to know what to do when.  The more variables attached to an AI decision tree, the higher the likelihood we will always find something that really messes with your game.

10 hours ago, Xorg_Xalargsky said:

I've done simple tests in the past and troops will exit trenches placed in the open. It also seems this is something you see mostly with infantry occupying fortifications against off-map artillery.

I have tested against on map mortars and the behavior is pretty much the same.

Link to comment
Share on other sites

12 hours ago, akd said:

2. Engine 4 problem that exacerbates the above: AI in cover will proactively displace from cover after receiving very light casualties.

It is not even required that the infantry suffers a casualty before they attempt to displace. The primary trigger of the flight response is a shell landing in close proximity. This is best showcased in the video when the German MG-42 team set to max stats packs up and runs the moment a shell lands near it. They don't even take a casualty before doing this. 

 

12 hours ago, akd said:

Solution: instead of AI recognizing indirect / HE threat and proactively displacing, AI should recognize foxholes / trenches as good cover versus indirect / HE and proactively use this cover when subjected to indirect / HE (cower / hide to minimize exposure).  Artillery vs. AI in cover then becomes more about suppression than destruction (as is the case in reality).

I tend to agree with your end results here as a solution, but I do not think a complete reworking of the TacAI is necessary. The TacAI in v3 was fine. The new behavior is supposed to allow the TacAI to displace when under threat of direct HE (example: not standing still in the middle of the open as they get blasted by a tank) or to seek cover if caught in the open during an artillery barrage. (example: dashing into a nearby building/fortification if in the open when shells start landing) The current problem is a problem because the displacing behavior triggers regardless of whether or not the infantry is already in cover. My solution would be to make it so infantry only displace if caught in the open. They would not displace if already in good cover. 

 

11 hours ago, Bulletpoint said:

Actually trees protect quite well against artillery in this game.

Please stay on topic. This thread is about how the 4.0 TacAI reacts to incoming HE (both direct and indirect) and how there could be a flaw/bug in the displacement logic. It is not about the performance of artillery or about the effects of tree burst. Start another thread to discuss this if you feel the need, but I would ask you do not continue to discuss it here. 

9 hours ago, Sgt.Squarehead said:

Looks pretty conclusive doesn't it.  :unsure:

I agree, however I want to stress that I do not consider this video to be a conclusive test. It is only meant to be a showcase of readily observable behavior in 4.0.

7 hours ago, sburke said:

I suspect there isn't an easy answer for the AI in single player mode to know what to do when.  The more variables attached to an AI decision tree, the higher the likelihood we will always find something that really messes with your game.

This is true. I very much want to avoid the rabbit hole of debating the TacAI in general. That is not my intention here. What I am trying to point out is that the TacAI behavior in v3 is overall better when it comes to how they react to indirect fires. Again, the improvement to the TacAI in 4.0 is supposed to allow the TacAI to automatically displace if they are getting blasted away in the open. If this was what actually occurred in game, there would be no issue here. The problem is that while the TacAI will displace after getting blasted in the open, they also run when they are already in good cover. This is the primary issue. 

Link to comment
Share on other sites

3 hours ago, IICptMillerII said:

I agree, however I want to stress that I do not consider this video to be a conclusive test. It is only meant to be a showcase of readily observable behavior in 4.0.

true to a degree, but I was seeing some similar behavior in MG in what I think was then 3.0.  I watched squads of guys start bailing out of a building when 82mm mortars were dropping on the roof.  They'd run outside and then start dropping like flies from mortar rounds.

Link to comment
Share on other sites

On 2017-08-13 at 1:24 PM, Rinaldi said:

I've sent @sburke saved files showing this problem and can corroborate - in fact, Miller was my playtesting buddy in those saved files. He hasn't pulled this out of nowhere, I assure you. We've been mulling over how to articulate the issue without getting cornered by ifs and whats. I'm confident enough now in saying that this isn't, as is often the case, players interpreting combat differently from the devs.

Then thank you both - this is helpful.

15 hours ago, IICptMillerII said:

This is true. I very much want to avoid the rabbit hole of debating the TacAI in general. That is not my intention here. What I am trying to point out is that the TacAI behavior in v3 is overall better when it comes to how they react to indirect fires. Again, the improvement to the TacAI in 4.0 is supposed to allow the TacAI to automatically displace if they are getting blasted away in the open. If this was what actually occurred in game, there would be no issue here. The problem is that while the TacAI will displace after getting blasted in the open, they also run when they are already in good cover. This is the primary issue. 

Nice summary.

As @sburke said these are very likely side effects of the desired change to have troops caught in the open under direct or indirect HE fire should displace to avoid just staying put and becoming casulties.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Unfortunately, your content contains terms that we do not allow. Please edit your content to remove the highlighted words below.
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...