A microscopic agent formulation is an appealing approach from a simulation perspective for many complex systems involving cooperative behaviour. It is satisfying to construct a detailed localised model of contributing agents and to experiment with a collective to study emergent effects in the overall system without having to build in global heuristics that anticipate known solutions or behaviours. The collective world in which the agents transact their operations can take several different forms, the most general of which is an arbitrary graph. We describe our simulation framework engine for studying cooperative effects amongst agents on graph structures and report on some experiments on path-finder agents that are limited to localised knowledge and heuristics. We explore some consequences of graph connectivity and the interplay between short and long-range agent spatial knowledge and present some preliminary results on autonomous exploration agents. We also describe ideas and issues for generalised simulation engines for interacting agents on graphs.
Keywords: agent; graph; simulation engine; simulation visualisation; path-finder experiments.
Full Document Text: PDF version.
Citation Information: in Proc. Int. Conf on Modeling, Simulation and Visualization Methods, (MSV'06), Las Vegas, June 26-29 2006.
BiBTeX reference:
@inproceedings{CSTN-030,
address="Las Vegas, Nevada, USA",
title="Simulating Cooperating Localised Agents on Graphs",
author="K.A.Hawick and H.A.James",
booktitle="The 2006 International Conference on Modeling, Simulation & Visualization Methods",
editor="H.R.Arabnia",
year="2006",
pages="CD",
month="June",
note="CSREA Press, ISBN 1-932415-99-8",
series="CSTN-030"
}
\bibitem{CSTN-030}
Simulating Cooperating Localised Agents on Graphs,
K.A.Hawick and H.A.James,,
Edited by H.R.Arabnia, Las Vegas, Nevada, USA, 26-29 June 2006, CSREA Press, ISBN 1-932415-99-8,
and Technical Note CSTN-030.