Will CMx2 ever have "strategic AI" ?

[Pandur]

well, we all know that CMx2 has tactical AI(in short tacAI) that works reasonable well in most situations but when it comes to strategic AI(in short stratAI) the game falls horribly short. in fact there is no stratAI to speak of.

now nearly 4!!! years after the release of CMSF, the ways to mimic stratAI are still the same inadequate tools, in short there is only the "clock". troops wont move unless the clock reaches a certain time or the tacAI tells the unit to run away because of incoming firepower. but no unit will ever go forward on its own decision unless the tacAI messes things up and retreats a unit towards the enemy.

there has been much talk about more triggers and conditions that can be used to more effectively scrip the computer players behavior and this discussions came up rather soon after the release of CMSF. still nothing got better in this regard. simple triggers with conditions could help tremendously into making the player believe that there is some sort of stratAI at work that reacts to anything but the clock.

i am pretty sure that steve said back in 07 or 08 that the military wanted to have scriptable behavior for enemy units, also to have no stratAI at all is a big time saver in development, but now that the scriptable behavior did not get the military contract that was hoped for, does that mean it wont be revisited and actually made more functional, but the concept is ditched instead??? is the "unit moves by clock" behavior all that will ever be in CMx2? what about more AI groups then 8, at least?

i know, the autonomous AI in CMx1 had its sometimes big shortcomings but after playing CMx2 for quiet some years now i rather wish back real AI or at least more(a lot more)scripting functions.

in my honest opinion, for a game that is focused on single player the AI or lack of it, is greatly hampering the single player aspect of the game after you played it for a week or two and "got" what its all about.

what, maybe for CMx3 you say?

[Ranger33]

I would love to at least see triggers. Putting everything on a clock and then hoping the player advances at a certain rate to match that clock makes for uninteresting battles. The framework is already there with the detailed objective system where individual areas and units can be "targeted". Now just allow the scenario designer to connect those conditions with movement orders and reinforcements, and the game would really open up.

[Alan8325]

I would love to at least see triggers. Putting everything on a clock and then hoping the player advances at a certain rate to match that clock makes for uninteresting battles. The framework is already there with the detailed objective system where individual areas and units can be "targeted". Now just allow the scenario designer to connect those conditions with movement orders and reinforcements, and the game would really open up.

+1, as long as those triggers are based on spotting. If I move my troops somewhere and it causes the AI to react without them actually seeing me there, well that's cheating.

[Georgie]

I agree, making the defense depend upon a clock doesn't work too well. I am a slow "advancer" so the clock runs out and the defense does some stupid things like moving a panther that had a perfect position to the rear where it is out of sight and in a much worse position for the defense. BF has so much on its plate to do that this problem is probably way down at the bottom of the to do list and that's too bad because its key to the long term playability of the game.

[LemuelG]

Be careful what you wish for. There are many a fine prospective tactical engagement over the years which 'AI' promptly ruined within seconds. I would much rather see the current plan system fleshed-out more.

[Ranger33]

+1, as long as those triggers are based on spotting. If I move my troops somewhere and it causes the AI to react without them actually seeing me there, well that's cheating.

Isn't spotting certain units already a possible objective? Would work perfectly.

Just imagine this: A scenario could be designed so that if the AI loses an entire platoon of tanks crossing a certain area, the next platoon could be triggered to take a different route! Brilliant!

Be careful what you wish for. There are many a fine prospective tactical engagement over the years which 'AI' promptly ruined within seconds. I would much rather see the current plan system fleshed-out more.

Fleshing-out is exactly what this would be. Just building on what is already there to make the AI actually able to react to the situation, instead of mindlessly trudging on along it's predetermined path, regardless of events.

[LJFHutch]

What I'd really like to see is better management of the forces. Too often I've come across panthers sitting behind high bocage doing nothing, if they had been moved 20m along to the low bocage they could have destroyed my advancing forces. It's not just a QB issue either, although in QB's the plans have to be much more general.

[Paper Tiger]

I'm not sure how a new level of Strat AI could interact with the current system without breaking it. Do you mean move towards VP locations of its own volition or when they change hands? I play exclusively against the AI and I find the current timing scripting, along with the fact that we can have up to five AI plans for any mission can present me with a far greater challenge than the old CMx1 AI could. The last point alone is the most important for me as it means I can replay missions I designed myself without me knowing what the AI is going to do or where its units are at set-up so as I can blast them into oblivion with my pre-planned artillery barrage.

Now, triggers will go a long way to giving us what we all want. I can't wait to get started work with them when they finally arrive. The current system could be substantially improved as well simply by allowing us to use 16 or 24 or whatever, AI groups instead of the eight we are presently restricted to using. For example, you have a lot of control over what an AI group will do when your AI group consists of only one vehicle.

I will add that I am in the middle of playtesting a new mission and I've seen the Tac AI do some pretty impressive things that I didn't know it could do, like reposition an AT gun to fire on forces that are out of its LOS. We're not talking about swivelling in place either

When changes do come, they will be in a new title and so I don't expect we'll see them in CMBN, or CMSF either. I think we're stuck with what we've got for the life of CMBN as any change might break existing work.

[sburke]

I will add that I am in the middle of playtesting a new mission and I've seen the Tac AI do some pretty impressive things that I didn't know it could do, like reposition an AT gun to fire on forces that are out of its LOS. We're not talking about swivelling in place either

That is most impressive, kind of scary too. I haven't even managed to do that successfully.... I suck. Friggin AI trying to make me look bad.

[LongLeftFlank]

Be careful what you wish for. There are many a fine prospective tactical engagement over the years which 'AI' promptly ruined within seconds. I would much rather see the current plan system fleshed-out more.

To my mind, a CMx2 compatible trigger (which is the entry of a designated zone by an enemy [choose one] any unit / infantry / AFV) would simply act as an EXIT AFTER command for the designated AI Group, at which moment it would execute its next Order. That Order doesn't have to be "everybody move directly to that zone right away and duke it out". Until the triggering condition is met, the AI Group doesn't execute that Order or any succeeding Orders.

This is one reason why the game needs at least 20 AI Groups, not 8.

[MikeyD]

Does Pandur play the game? He's making it sound as though AI behavior in 2011 is the same as AI behavior in 2008.

[Paper Tiger]

I'll let Pandur answer that one but I will say that he is a very good scenario designer and I can understand his frustration. Event triggers as opposed to time triggers are what most of us want and that will make him a happy camper again.

[hoolaman]

Does Pandur play the game? He's making it sound as though AI behavior in 2011 is the same as AI behavior in 2008.

Man what is it with the constant snide one liners? Of course he plays the game, and the Strat AI from 2011 IS the same as the one from 2008.

[Pandur]

as hoolaman said, it IS the same. ways to mimic stratAI did not change in all that years. and yet again i am suprised by the lack of understanding of the game system by some beta testers.

still, only the exit after or befor conditions and 8 AI groups present, the same was in CMSF in 2007 actually.

i do not talk about tacAI which got worked on and got better over time.

Event triggers as opposed to time triggers are what most of us want and that will make him a happy camper again.

i could not have said it better myself. maybe the thread title is a bit misleading in regard to what i want. i do NOT expect to suddenly be presented with an autonomous AI but wish for a lot more ways to mimic autonomous AI. especially since i like large scenarios, it is extreme guesswork on the scen designers side where the player will be after like 120 turns or more, and set the AI script accordingly to what he thinks could happen after 2 hours of gameplay. i can observe on myself that this leads to very much simplified AI plans that should work on any situation in the mind of the scenario designer.

@Paper Tiger

i am a very good scenario designer? wow that is nice to hear, i barely made 2 scenarios, where one admittedly took a looong time to get finished, but i would not have called myself a very good scenario designer. nice to hear that from someone else