From: Morimoto Takeshi (morimoto@takopen.cs.uec.ac.jp)
Date: Tue 15 Oct 2002 - 02:57:41 GMT
Dear all,
We had to mention the background of our two proposals about interagent
communication more clearly:
(1) the first proposal in [rescue:5506] and
(2) the counter proposal in [rescue:5534].
At first, our major premise is that an agent is a human or a group of
humans in the current RoboCupRescue Simulation System (RCRSS).
Because a human cannot speak and hear enormous amount of messages, it
is natural that an agent also cannot do so.
However, the current rule does not restrict the length of messages for
interagent communication. Our proposals aim to model a flood of
information in the RCRSS: an agent has to select messages from numbers
of received ones; and an agent has occasionally to restrain speaking
messages to prevent the flood.
To model the information flood produced by the agents in the field,
our proposals have:
- a rule to restrict the length of a message and
- a rule for an agent to be able to select messages it wishes to
hear from all received messages,
and these rules do not need to change the current kernel code.
If many people think the proposal (1) is too restrictive, we do not
claim that it is better than the late proposal (2). We do not believe
that both proposals deteriorate the interest of RoboCupRescue as a
good distributed multi-agent problem where agents should colaborate in
quite a bad circumstance.
In the future, we shall have to introduce such as the agent
communication model proposed by [I. Noda et al., Language Design for
Rescue Agents, pp.164--172, RoboCup2001: Robot Soccer World Cup V
(LNAI2377), Springer Verlag, 2002] in the RCRSS, and agents will be
able to communicate with each other in a rich protocol.
Regards,
-- Prof. Ikuo Takeuchi, Takeshi Morimoto, Hiroki Endo The University of Electro-Communications, Japan
This archive was generated by hypermail 2.1.4 : Tue 15 Oct 2002 - 03:24:29 GMT