Newsgroups: comp.ai.games
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!nntp.sei.cmu.edu!cis.ohio-state.edu!math.ohio-state.edu!howland.reston.ans.net!newsfeed.internetmci.com!news.mathworks.com!news.kei.com!wang!news
From: bruck@actcom.co.il (Uri Bruck)
Subject: Re: Town Simulation
Organization: ACTCOM - Internet Services in Israel
Date: Tue, 17 Oct 1995 23:11:35 GMT
Message-ID: <DGM8FC.5GM@actcom.co.il>
References: <44rvgm$q2p@rc1.vub.ac.be> <44vdp1$dno@spectator.cris.com> <454p7d$9sv@linda.teleport.com> <1995Oct9.171252.10713@llyene.jpl.nasa.gov> <45o231$m8e@mhost.ida.his.se>
Sender: news@wang.com
Lines: 25

Kim Laurio (dv3kimla) wrote:
: How do the agents select an action?
: How will the agents make plans/decisions? (search algorithms)
: What will/can the agents sense of the surrounding world?
: How will they sense it?
: How do you write down the "laws" of this world? Do you wan't the agents to just
: know from start (a lot of coding in every agent), or do you wan't them to be
: expressed in a database = outside the agent?

These two options are not really contradictory to each other. You can provide
an agent with a database (wouldn't a knowledge base be a better term), and
decide how much information you want to provide it with. With
many agents, you can start each one off with access to different inamounts
of information

: Well, there are a lot of questions here to be answered and I suggest that you
: start VERY small and proceed in a VERY careful/slow pace to try different
: models
: and evaluate their strenghts/weaknesses. Like a 4*4 world with a very limited
: set of objects to interact with and some simple ways to act. I promise you,
: that
: will be hard enough to start with...

Probably will be

