Re: Computer version of Risk
> > > The AI should be pretty straightforward.
> >
> > Hope so. I'm new to AI.
So am I, so don't accept my word as law...
> All AI programming is centered around 2 concepts:
> 1) What does the computer player know
> 2) What does he do about what he knows
Ooh.. That's a very simplistic view...
> If you ask 5 risk players how to play, you'll get at least 6 "optimal"
> strategies.
----------8<----------
strategy stuff deleted
-------->8------------
> The "best" way to create computer AI is to make the game playable by human
> players, then play about 20 games with human players and have the computer
> keep track of every move of every player. You then can "teach" the
> computer to play like you do.
And who says you're so good? (Don't take it personally, but I'm not convinced)
----------8<----------
more stuff deleted
-------->8------------
The normal aproach to board-game AI's is the one (normally) used for chess:
Max-Min algorithm
=================
Make a function to give each board-situation a value (score if you like).
1. Make a tree of all possible moves (except suicidal ones), and evaluate them
(the resulting board), and cut away the most stupid moves. (lowest score)
2. For every considered move on your part, (the tree) make a move-tree for the
opponent. Assume that he is clever too, and won't make any obviously stupid
mistakes.. (Store these moves with the score they would give you)
3. Repeat 1&2 either to some specified depth (height of tree) or to some other
limit...
4. Find the leaf with the highest score (for you), and make the first move in
that
direction...
It's only in real-time games or games with *enormous* branching that really
require something as obscure and intagible as "tactics and strategy". For Risk,
I'd write the valuation function to valuate for any player, since *each* player
is trying to *max* his own score. (If there were trying to *minimize* *your*
score, like chess AI's, they'd all gang up on you!
)
Hope it helps... I'll try to get some work in on the tree.e...
(Hey! Here's another use for trees!)
>>Joseph Martin
> Michael Packard <lgp at exo.com>
Anders
-------------------------------------------------------------------
Anders Eurenius <c96aes at cs.umu.se> ICQ UIN:1453793
Computer Science/Engineering student at the university of Umeaa
-------------------------------------------------------------------
|
Not Categorized, Please Help
|
|