Vidal's libraryTitle: | Adaptive Robot Coordination using Interference Metrics |
Author: | Avi Rosenfeld, Gal A. Kaminka, and Sarit Kraus |
Book Tittle: | Proceedings of the AAMAS-04 Workshop on Learning and Evolution in Agent Based Systems |
Year: | 2004 |
Abstract: | One key issue facing robotic teams is effective coordination mechanisms. Many robotic groups operate within domains where restrictions such as limiting areas of operation are liable to cause spatial conflicts between robots. Our previous work proposed a measure of coordination, interference, that measured the total time robots dealt with resolving such conflicts. We found that a robotic group s productivity was negatively correlated with interference: Effective coordination techniques minimized interference and thus achieved higher productivity. This paper uses this result to create adaptive coordination techniques that are able to dynamically adjust the efforts spent on coordination to match the number of perceived coordination conflicts in a group. Our robots independently calculate a projected level of interference they will encounter. By using this metric as a guide, we are able to create adaptive coordination methods that can quickly and effectively adjust to a given domain s spatial limitations. We present two adaptation heuristics that are completely distributed and require no communication between robots. Using thousands of simulated trials, we found that groups using these approaches achieved a statistically significant improvement in productivity over non-adaptive coordination methods. |
Cited by 5 - Google Scholar
@InProceedings{rosenfeld04a,
author = {Avi Rosenfeld and Gal A. Kaminka and Sarit Kraus},
title = {Adaptive Robot Coordination using Interference
Metrics},
booktitle = {Proceedings of the {AAMAS-04} Workshop on Learning
and Evolution in Agent Based Systems},
year = 2004,
abstract = {One key issue facing robotic teams is effective
coordination mechanisms. Many robotic groups operate
within domains where restrictions such as limiting
areas of operation are liable to cause spatial
conflicts between robots. Our previous work proposed
a measure of coordination, interference, that
measured the total time robots dealt with resolving
such conflicts. We found that a robotic group s
productivity was negatively correlated with
interference: Effective coordination techniques
minimized interference and thus achieved higher
productivity. This paper uses this result to create
adaptive coordination techniques that are able to
dynamically adjust the efforts spent on coordination
to match the number of perceived coordination
conflicts in a group. Our robots independently
calculate a projected level of interference they
will encounter. By using this metric as a guide, we
are able to create adaptive coordination methods
that can quickly and effectively adjust to a given
domain s spatial limitations. We present two
adaptation heuristics that are completely
distributed and require no communication between
robots. Using thousands of simulated trials, we
found that groups using these approaches achieved a
statistically significant improvement in
productivity over non-adaptive coordination
methods.},
keywords = {learning robots},
url = {http://jmvidal.cse.sc.edu/library/rosenfeld04a.pdf},
comment = {masrg},
cluster = {6099933472055950917}
}
Last modified: Wed Mar 9 10:16:12 EST 2011