Jump to content

Will AI "scripts" continue to be the norm for CM games?


Recommended Posts

Sorry, I have to dispute that. "completely unflexible" is not the term to describe the AI. The AI is given a script yes, but TAC AI takes over and makes it work. Any battle is forged from a plan. Real life or not.

Plus a designer can implement several plans with several groupings. You can have about 100 different combinations. It takes some imagination, planning and analytical process. A grasp of tactics also helps.

The AI of CMX1 was adequate but not fluid. It saw the Flags and went for them. 100% predictable. Hardly ever a flanking move or anything fresh. I am not being mean but I think your approach to "making it better" is flawed. The CMSF AI and future developments to the AI will make the first attempt at AI look stoneage. I would rather have the AI as it is than take a massive leap backward to the Zerg rush we had before.

Link to comment
Share on other sites

Cerebrus, I beg to differ. With the old AI it was pretty boring to play anything but attack. When you played as the defender it was only interesting by giving the AI overwhelming odds. And besides BFC have already stated that it would be impossible (or too hard to be worthwhile) for them to make a credible AI for modern combat given the higher variety of weaponry and capabilities.

What I would like to see though, are greatly expanded "scripting" tools, such as triggers for plans to move to the next step, allow for sub-plans and branching between them based on enemy and friends dispositions, condition, ammo, etc. I would like to see AI triggers and general scripting functionality for a proper fire support plan, I would like to see the designer able to not only specify that he wants a force to go to a specific location, but also to "point out" to the AI a preferrable route or routes to "consider" based on known enemy dispositions. I would like to see the designer being able to coordinate different forces' plan steps so that e.g. force A only moves to step 3 when force B moves to step 4, but will only wait up to 5 minutes before not waiting any longer, etc.

Given the amount of expertise, knowledge, ingenuity and dedication of the community, even some of these tools would result in absolutely massively, clever, diabolical, surprising, exciting, ingenious scenarios. Consider the limited tools we have now and compare with the wonderful scenarios and campaigns already on offer, they are way better than CMx1 scenarios in my opinion.

Cheers

Link to comment
Share on other sites

The new engine is way to complex to turn the enemy StratAI into a generic script like it was with the CMx1 games. There are more objective types, more complex surroundings, and the engine will span more than just one time period. Scripting is here to stay.

What will improve over time is the ability for the scenario designer to put together useful scripts. Personally I'm hoping for the addition of "triggers" as soon as possible, but other features would be copying of AI plans so you can create even more variations quickly for example.

Link to comment
Share on other sites

What will improve over time is the ability for the scenario designer to put together useful scripts. Personally I'm hoping for the addition of "triggers" as soon as possible, but other features would be copying of AI plans so you can create even more variations quickly for example.

These are two biggies for me, especially copying. It is incredibly tedious as it is.

Other AI things that would be nice to see:

- Ability to name AI plans.

- Ability to switch off AI plan 1 (As in "not used"). It's kinda annoying when trying to test the other AI plans by themselves.

- More than 8 AI groups.\

- A better explanation of how the "exit before/after" works. I've experimented with it, and it STILL confuses me.

- Better dismount behavior. It's hard for me to get the AI to dismount without sitting around for ten minutes on the objective drawing enemy fire.

- A on/off switch for individual AI units in a group to regroup at each waypoint. If it is off, the AI unit immediately proceeds to the next waypoint without this stop-and-go movement they do now.

- Ability to dictate a time for AI artillery to begin a fire mission. Useful for getting it to use smoke screens correctly, since timing is everything. Also, ability to define particular AI artillery batteries/sections to particular targets.

- Ability to define target type, so that ATGMs don't expose their position to a measly humvee, for example.

Or that, or we could just go back to having the AI crawl across the map in a big blob on permanent "hunt" mode. :D

Link to comment
Share on other sites

Biggest problem in current StratAI is it's inability to react to anything. This is what CMx1 AI did better (maybe i sucked at playing, but it gave me supprise or two with it's counterattacks)

That makes it bad-bad defender. All CMSF missions with AI as defender are static bughunt (doesn't anyone see? am i the Oracle? Do i have the Sixth sense?). They sit in their holes and suck it. Reserves are just those troops who suck in their holes bit farter away in Setup.

It's far too passive, if it loses objective right under it's nose it just sits in it's hole and suck it. If there is break thru in south, it's T-90s/M1A2s are staying nothern side of hill from where one isn't comming (well... airstrike is!).

So i think somekind reaction plan should be inside AI plan (or something).

-What objectives is vital to keep

-what units are primary reserves, which are not to leave their positions under any condition.

-more...

Sure it must take quite much time to push this into code. But... Problems with CMSF seems to return to this point. This is bigger of two most gamebreaking thinges to me.

Let's keep in mind boys and girls: No plan lasts with first contact with enemy. That is why i always run on amok without plan :D

Link to comment
Share on other sites

Biggest problem in current StratAI is it's inability to react to anything. This is what CMx1 AI did better (maybe i sucked at playing, but it gave me supprise or two with it's counterattacks)

That makes it bad-bad defender. All CMSF missions with AI as defender are static bughunt (doesn't anyone see? am i the Oracle? Do i have the Sixth sense?). They sit in their holes and suck it. Reserves are just those troops who suck in their holes bit farter away in Setup.

It's far too passive, if it loses objective right under it's nose it just sits in it's hole and suck it. If there is break thru in south, it's T-90s/M1A2s are staying nothern side of hill from where one isn't comming (well... airstrike is!).

So i think somekind reaction plan should be inside AI plan (or something).

-What objectives is vital to keep

-what units are primary reserves, which are not to leave their positions under any condition.

-more...

Sure it must take quite much time to push this into code. But... Problems with CMSF seems to return to this point. This is bigger of two most gamebreaking thinges to me.

Let's keep in mind boys and girls: No plan lasts with first contact with enemy. That is why i always run on amok without plan :D

This is a really, really good point and criticism. I think it could be solved with scripting and triggers, although the scenario designer then has to think in a "what-if" fashion: "If enemy captures Objective A, send reserves on this plan".

Link to comment
Share on other sites

Normal Dude. Good points...

Other AI things that would be nice to see:

- Ability to name AI plans.

I would LOVE to see that too. I often name my plans on paper and it would be nice to have this feature to remind us what each plan is designed to do as well.

- Ability to switch off AI plan 1 (As in "not used"). It's kinda annoying when trying to test the other AI plans by themselves.

This is on the to-do list. I believe it was Moon who told me that it might make the Brit module. It is SO frustrating setting Plan 1 to use rarely to playtest a new AI plan but after 20-30 minutes playtesting, you notice that you're playing plan 1 AGAIN!!!:rolleyes:

- More than 8 AI groups.\

Absolutely. I think this one change alone would improve the quality of the AI in missions exponentially. I'm sure you meant to include more than 16 orders per plan too as without extra orders, those extra groups would just do squat. It's been brought up before but it didn't get noticed by BFC.

- A better explanation of how the "exit before/after" works. I've experimented with it, and it STILL confuses me.

Ha ha. I'm glad you 'fessed up' first. I have to confess that I'm still a bit vague about it too. Since this has been the subject of more than a couple of threads, it's remarkable how it has been ignored by BFC. Steve is more than happy to spend hours of keyboard time explaining other, frankly, less important aspects of the game to people who don't understand them. And I was disappointed to see that the new V1.10 manual hadn't received an update. I suspect that they don't know how it works either.:D

No comments on your next two points.

- Ability to dictate a time for AI artillery to begin a fire mission. Useful for getting it to use smoke screens correctly, since timing is everything. Also, ability to define particular AI artillery batteries/sections to particular targets.

Yes, a nice optional delay for the support target zone would be fantastic, just like the player has when he is making his pre-planned artillery barrage. Immediate or 5 minutes sir? It would make AI artillery even more effective and keep the AI honest too.

I'd also like to have AI artillery support plans integrated into an AI attack plan so that the artilery hits/interdicts those areas that are important to that particular plan. I'd also like a Porche.

Link to comment
Share on other sites

Secondbrooks

Biggest problem in current StratAI is it's inability to react to anything. This is what CMx1 AI did better (maybe i sucked at playing, but it gave me supprise or two with it's counterattacks)

Agree with the first part but definitely not the second. I used to get frustrated that the AI would just abandon a nice defensive position to go and counterattack some pokey little flag, often suicidally and with no effect. As a result, the flag they were supposed to be protecting was much more vulnerable.

Sure, the total non-responsiveness of the AI to the player's actions during a mission is a pinch but I believe that BFC are working to include scripting into the AI programming at some point in the future. In fact Moon almost said as much earlier in this thread. No doubt, it will enable the AI to react to players actions more effectively during a mission. Then we'll all be happy :D

Link to comment
Share on other sites

I vote for scripting too. Its far better. We just need more tools to handle our plans. I usually use some vehicles with some units together, it is working fine, and I like how they cooperate. Now in editor we are painting the zone with yellow color where the group has to move - works fine, but for better result some precision in it will be needed. We have two houses, every on other side of the road. We paint the houses and the road between – units enter mostly the houses and vehicle stays on the road, but sometimes also some units stays on the road...

Could it be possible to have for example possibility to paint with dark yellow place where only vehicles can go and with brown places where only infantry can go? Usual yellow will mean both can go there. It will help for exact placing of vehicles without unit especially now, when there is this secondary explosion effect, which can ruin quite lot of nearby soldiers....

Link to comment
Share on other sites

That non reative AI problem would be solved by adding in conditions, events and triggers.

The Ai has a unit guarding a buiding, you write a condition for the unit: casualties = 50% to 100% randomised. You have the backup unit and give it an event of going to reinforce the building if the first condition is true (with a 80% chance of doing it for randomness sake). Then you have a trigger for the reinforcement unit: if it detects enemy units under x distance 40% of returning to origional defensive line. If it recieves fire from enemy unit 80% chance of returning.

A very simple script but it adds great dynamics to the battle, imagine what 100s of complex scripts could do?

Link to comment
Share on other sites

I think Steve had mention a hope to eventually get some sort of 'triggers' into the game but you can imagine the amount of coding and re-laying out of the editor that would involve. Every bright idea we players think of we're laying at the doorstep of the upcoming WWII title - expecting to see triggers or water or star shells or a dozen other things more difficult to code than the next. I wonder what their priority list looks like, most of the time AI improvements have stayed at the very top.

Link to comment
Share on other sites

The TacAI in CMx2 can be as flexible or as tight as the scenario designer makes it. Too many orders, with too small destination areas results in a computer opponent that does not adapt at all to the player's actions.

I'd appreciate if Battlefront would make triggers available, though.

Cheers,

Link to comment
Share on other sites

That non reative AI problem would be solved by adding in conditions, events and triggers.

The Ai has a unit guarding a buiding, you write a condition for the unit: casualties = 50% to 100% randomised. You have the backup unit and give it an event of going to reinforce the building if the first condition is true (with a 80% chance of doing it for randomness sake). Then you have a trigger for the reinforcement unit: if it detects enemy units under x distance 40% of returning to origional defensive line. If it recieves fire from enemy unit 80% chance of returning.

A very simple script but it adds great dynamics to the battle, imagine what 100s of complex scripts could do?

This is exactly what would be great to have. Those AI groups could send signals that some other groups could react to. You could for example first send a small group to some part of the map. If they get there with casualties below certain level, they'd trigger 'road is clear' event and another AI group waiting for this event could follow.

Link to comment
Share on other sites

This is a really, really good point and criticism. I think it could be solved with scripting and triggers, although the scenario designer then has to think in a "what-if" fashion: "If enemy captures Objective A, send reserves on this plan".

Thank you. :cool:

Pretty much like that was my toughts as well. Scenario designer is best one to think abotu these things, that way we can make opponents without zero will to react or then some duracell-bunnies who do mad-flag-rushes instantly. CMx2 community has proven in my mind it's ability in desigining good AI plans for specific missions.

Paper Tiger: Yes. The way CMSF has AI plans has proven to be very good idea, much better than forexaple waypoints in other games has. I'm not against it, just wishing that some reactivity can be added into it. My computer hasn't work ever well with CMx1s so i dont' have great understandment on it's AI, but yeah i remeber seeing some truly dumb choises.

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...