🚀 go-pugleaf

In article <Dec30.183952.57153@yuma.ACNS.ColoState.EDU>, hallch@CS.ColoState.EDU (christopher hall) writes:
> Hello all,
> 
>      I was wondering if anyone could point me in the direction of any
> material related to the use of artificial intelligence in computer games.
> Specifically I am searching for material that would be useful in
> controlling computer players in a role-playing game.  I am assuming that I
> will have to program the different possible actions for the computer but I
> would like to know a good and efficient way to go about making the
> computer decide which option would be the 'best'.

You can get some rather nice behaviors out of a rule based system.  First,
you define all the actions your beasties need to perform.  Second, define
a list of things they can know, such as "is there a person in front of me"
and "I am really, really hungry".  The knowlege base should be true/false
items so they are easier to compute.

Then create a list of actions.  The beastie will perform the first one where
his current state matches the required state.  The trick to make this 
useful, is that your state consists of a mask showing which of the t/f items
are important, and another variable showing the required state for a specific
action.  That way, you can do things like the following with the rules:

if(hungry AND near food AND NOT near enemy) then eat the nearest food.

This technique also works well in things like AI players for military
simulations, or anything where you have multiple non-communicating opponents.
If you're really sneaky, you can also add communicating opponents. :^)

John Olsen
olsen_jo@uvadm.uvcc.edu