@COMMENT This file was generated by bib2html.pl version 0.94
@COMMENT written by Patrick Riley
@COMMENT This file came from Gal A. Kaminka's publication pages at
@COMMENT http://www.cs.biu.ac.il/~galk/publications/
@PhdThesis{rosenfeld-phd,
author = {Avi Rosenfeld},
title = {A Study of Dynamic Coordination Mechanisms},
school = {{B}ar {I}lan {U}niversity},
year = {2007},
OPTkey = {},
OPTtype = {},
OPTaddress = {},
OPTmonth = {},
OPTnote = {},
abstract = {
Coordination, or the act of managing interdependencies
between activities, is a key issue within the field of multi-agent
systems. Because of the importance of this issue, many theoretical
and practical frameworks have been proposed for addressing
coordination challenges. However, finding the optimal coordination
method for a given a group of agents a domain task is a
computationally difficult, if not intractable, problem in most
real-world domains. Solving the coordination problem is thus an
important open challenge for researchers in this field.
Towards addressing this issue, this thesis presents an algorithm
selection approach for creating adaptive coordination methods. We
study several types of coordination problems from robotic foraging
and search domains, constraint satisfaction and optimization
domains, and Peer to Peer networks. We find that novel teamwork
measures can be developed for quantifying the effectiveness of
coordination algorithms in all of these domains. These measures can
be autonomously and locally measured by team members, even without
any communication. The significance of this result is its ability to
effectively quantify coordination in a clear, tractable fashion.
Next, we find that these measures can be used to switch between
coordination methods as needed. Robots or agents can effectively
select the best coordination method to their localized domain
conditions, online during task execution. The net result is a
significant productivity improvement of these adaptive methods over
the static methods they are based on.
},
wwwnote = {},
OPTannote = {}
}