[Fwd: Re: [Modeling] about modeling environment in agent-basedsimulation]

Danny Weyns danny.weyns@cs.kuleuven.ac.be
Sun, 29 Feb 2004 20:36:42 +0100 (MET)

Hello Jim,

the following paper describes a general architecture for situated
multi-agent systems in which environment is modelled as a first class entity:

"Formal Model for Situated Multi-Agent Systems", Danny Weyns and Tom
Holvoet, Formal Approaches for Multi-agent Systems, Special Issue of
Fundamenta Informaticae, Eds. B. Dunin-Keplicz, R. Verbrugge, 2003.

The formal part is based on set theory and is fairly understandable.

Another interesting paper that focusses on a model for environment that
deals with simultanoeus actions is:

"Model for Simultaneous Actions in Situated Multi-Agent Systems", Danny
Weyns and Tom Holvoet, First German Conference on Multi-Agent System
Technologies, MATES, Erfurt Germany, September 22-25, 2003, in LNCS Volume

Both papers are available on my website.

best regards,

> Danny,
> Are there any particular papers at your website that you could
recommend? Thanks.
> Cheers,
> Jim
> On 2/29/04 7:53 AM, Danny Weyns scribed:
>> Hello Ye,
>> your question concerns an important issue in agent-based systems. In my
opinion the environment need to be treated as a primary abstraction in
multi-agent systems, without any doubt!
>> In most models for mas, the environment is reduced to communication
infrastructure. However, this neglect the true potential of the
environment. Here are a couple of questions of my favourite list to
illustrate the necessity of the environment as an abstraction on its
own right:
>> -how to deal with objects or data in the environment
>> -what about the actions (possibly simultaneous actions) of agents onto
these objects
>> -how to control the effects of the agents' interactions independently
of the internal architecture of the agents
>> -how to deal with the (dynamic) topology of environments
>> -what about processes in the environment that do not directly
>> relate to agent activity such as the variation of an environmental
variable (e.g. temperature), the expansion of a field in the
>> environment, the movement of an object or the evaporation of a
>> pheromone?
>> We need the environment as an abstraction to model these issues, it seems
>> unnatural to me to put it all in the agents?
>> In my work, environment is treated as a primary abstraction, please
find my papers at
>> http://www.cs.kuleuven.ac.be/~danny/publications.html
>> Note that at the next AAMAS conference a workshop is dedicated to
environment for mult-agent systems, see
>> http://www.cs.kuleuven.ac.be/~distrinet/events/e4mas/
>> with regards,
>> Danny Weyns
> _______________________________________________
> Modeling mailing list
> Modeling@www.fipa.org
> http://fipa.org/mailman/listinfo/modeling