Suggestions for improving the AI artillery

[Paper Tiger]

Now that the game has largely been fixed, at least for those of us who don't play online, PBEM etc, I am hoping that some attention will be given to the condition of the AI's use of artillery in the game.

It seems there are a few other people who would like to see the AI be able to use artillery during the scenario instead of the 30 second delayed dump at the start. Here's how I think it could be implemented in the game.

We already have AI plans for the AI groups. First, each AI plan could have it's own stand alone AI artillery plan attached. Even with absolutely no other improvement to the system, that would help a LOT. But I suspect thsi won't happen any time soon.

Secondly, we could have the ability to create different types of artillery fire zones, an example being a) the preliminary opening barrage painted in yellow and then and in-game artillery barrage painted in red, (who cares!)that are ignored by the opening barrage.

That would allow us to add artillery as reinforcements later in the game and these modules could be fired at these 'red' artillery zones.

Now, that surely wouldn't require too much work to implement. I know that there a million and a half things that people want you to be spending your time developing but it really is about time the AI artillery got a bit of attention IMO only .

[Cpl Steiner]

Paper Tiger,

I know you are trying to go easy on BFC to make it more likely that a fix will be implemented but your approach treats artillery and air support as a Strat-AI issue, i.e. something the scenario designer needs to implement using the scenario AI tools.

In my opinion, this is the wrong approach. If you look at the American TOE, granted they do have FOs and JTACs who's job is specifically to call in artillery and air support. However, any unit in the TOE can call in this support as well, not just dedicated observers.

Given that US fire support is a tool in the arsenal of every single US unit, I think it has to be treated as a Tac-AI issue, not a Strat-AI one. In other words, just as the Tac-AI decides whether or not to open up with an M240B machine gun, so too should it decide whether or not to request an airstrike or artillery fire mission.

The choice of when to call it in, on what target, and what type of support to call, should be treated as similar to any other Tac-AI choices, such as what ammunition to select, which unit is the greater threat, ammo conservation etc.

I know this is more work for BFC but I personally would be dissatisfied with anything purely dependent on prebattle scripting.

[EarlofWarwick]

Just curious about this, so Battlefront please don't take this as a whining -- this is an authentic question--

What are the programming challenges for giving potential FOs the same or similar targeting capacities as tanks (i.e., the ability to designate a target and make the decision to fire).

We may be able to come up with better suggestions if we can better understand the difficulties involved.

Does the trickiness stem from the fact, for example, that FOs designate a patch of ground rather than a specific target?

VR

EoW

[YankeeDog]

FOs are capable of targetting a specific target such as a tank in CMSF, not just a patch of ground.

You don't have to be a genius in game programming to see that creating an AI capable of effectively using artillery is going to be quite a challenge. There's a lot of factors that come into play -- which unit should call in the strike, what kind of asset to use, etc. But I definitely agree that improvements in this area would advance the game considerably, especially when a human player is playing Red vs. an AI Blue force. Effective use of off-map support is critical to a realistic challenge from a Blue computer-controlled opponent.

Worth noting that the AI never used Artillery very effectively in CMx1, so this is really new ground here. the StratAI in CMx1 either never got around to dropping arty rounds at all, or dropped the whole FO load on the first unit it saw to wander into the spotter's LOS, or near a TRP. Occasionally, the AI would get lucky if you blundered your main effort onto a TRP or something, but usually it was ridiculously easy to get the AI to waste all its arty rounds on a small forward scout unit, or a defensive OP.

Probably a two-pronged approach is best, with the addition of tools to allow the scenario designer to map out fire plans, etc. for the AI *and* a gradual introduction of AI routines that allow the AI to actually do some fire support calls on its own.

It occurs to me that faster response support types (U.S. 81mm mortars for example) are probably a lot easier to program for elective AI use, than types with a longer response time, and/or more complex variables, like Air Support. Quick response stuff can be targeted based on current known location of enemy units. Slower response time stuff takes the ability to conjuecture where the enemy is *going* to be.

So perhaps BFC should start with, say, getting the AI to effectively use relatively simple support types like 81mm mortars, and some tools so that scenario designers can "help" the AI use arty effectively. Once they've got that down, they can build on that for the more complex support types.

Cheers,

YD

[MikeyD]

The trick would be to not cause more trouble than its worth. If the AI FO spots an enemy unit one building over then starts dropping mortars on his own position that would not be smart. AI FO targetting on-the-fly is probably easy to 'do' but it would be a real bear to 'do right'.

About the AI using artillery effectively, you should check out some of my own 3rd party scenarios. I love to see tons of artillery rocket fire rain from the sky. in one scenario I directly target the player's troops at startup (chuckle chuckle), in another I bring down a curtain of fire that impedes forward movement. Believe me, you dont want a scenario designer to REALLY crank up AI artillery assets. It would make for a very short game.

[YankeeDog]

Ya, that's why I think it would be best to take baby steps here. Start with the small stuff that's has a short response time and has a small danger-close radius.

Once the AT can drop 81mm mortar shells without hitting its own forces too often and at least occasionally causing damage to the enemy, then they can start worrying about adding routines that allow it to drop a 2000-lb. JDAM without wiping out half its own force. . .

[Paper Tiger]

Cpl Steiner, you're definitely right, I am trying to do this is a polite manner and keeping my tone reasonable. Trouble is, with reasonable and polite, you seem to have less chance of getting a response from BFC than if you attack the game. I suspect that somebody does read these threads but they are keeping such information out of the public domain. Unfortunately, I suspect AI artillery will be improved in the Marines module and not a patch so we might have a long wait for any improvements.

I love the idea that the TAC AI could handle artillery. Indeed, if you read the manual, the TacAI is supposed to do be able to do this already. It states that this function is beyond the scenario designers control and entirely used at the discretion of the TacAI. But that's obviously not working.

The system I outlined above would be very simple to use without having to write lots and lots of complex AI codes. It's almost identical to the system we already have only that the AI knows to fire on these 'Red' zones later in the game when the artillery reinforcements show up. Even such a simple system as this would be a MASSIVE improvement over the current system. It's very disappointing that the game doesn't allow the AI to fire artillery during a battle yet the human player can. I can fully appreciate that there were other issues that were far more pressing to fix before 1.07 but now, we're nearly there so this really needs some attention. There have been substantial improvements to the pathfinding routines, AI vehicle behaviour and even a whole new system of LOS introduced. Now, what about the artillery?