A more deliberate AI?

[blackcloud6]

To me, the AI advances fast and is very aggressive. This is good but it can cause the AI to get caught in kill zones, or overextended. What are some scenario design techniques to use to get the AI to move and fight more deliberately?

[CapnDarwin]

Can you give an example or two of the behavior you would like to see? We can work from there.

[blackcloud6]

Can you give an example or two of the behavior you would like to see? We can work from there.

Maybe have various "speeds" of approach (for lack of a better term.) Say when you draw a deliberate advance you could tell the unit to be aggressive (fast), moderate, or cautious (slow). maybe this exists already with orders/sop combinations.

Or allow you t time mission, such as advance to a point, and then tell the unit to hold for x number of minutes, then give another advance mission, then hold, etc.

[22sec]

It's really up to whomever is creating the BP's for a scenario, and has its own set of challenges the more complicated the plan becomes. I suppose not unlike real life. I've now created about ten scenarios with three BP's for each side, and I still am working on getting better. We will see what gets added to the scenario editor in the future.

[22sec]

[blackcloud6]

It's really up to whomever is creating the BP's for a scenario, and has its own set of challenges the more complicated the plan becomes. I suppose not unlike real life. I've now created about ten scenarios with three BP's for each side, and I still am working on getting better. We will see what gets added to the scenario editor in the future.

I was thinking about this today on my bike ride this morning. I guess the best way to learn how this works, is to do it. I need to experiment with subsequent orders and see how the units behave. What I want to achieve may already be in the game, I just need more practice.

[IronMikeGolf]

Necrothread, I know, but...

I think part of this is equipping the AI with a reactive capability, based on what it sees. Here's a very incomplete decision tree I am thinking of for an Advance Guard:

1. CRP makes contact
a. Enemy with no perceived anti-armor, CRP mows them down
b. Plt with anti-armor, call fires and observe
c. More than that, call fires and bypass

2. FSE
a. 1.a above, follow up on CRP
b. 1.b above, attack
c. 1.c above, engage to fix

3. AGMB
a. 2.a above, continue to march
b. 2.b above, bypass
c. 2.c above, bypass

This is the barest of skeletons. I think a useful approach is to define this in human terms and then see what we can implement in game code. I'd love to work through this with you, Fred.