Gal A. Kaminka: Publications

Sorted by DateClassified by Publication TypeClassified by TopicGrouped by Student (current)Grouped by Former Students

A Study of Dynamic Coordination Mechanisms

Avi Rosenfeld. A Study of Dynamic Coordination Mechanisms. Ph.D. Thesis, Bar Ilan University, 2007.

Download

[PDF]1.4MB  

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.

Additional Information

BibTeX

@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 = {} 
} 

Generated by bib2html.pl (written by Patrick Riley ) on Fri Aug 30, 2024 17:29:51