by Florian Klein, Holger Giese
Abstract:
Multi-agent systems present a promising paradigm for coping with the complexity of intelligent mechatronic applications. Frequently, they are characterized by the emergence of purposeful behavior and complex structures out of the seemingly chaotic interactions between their members. The safety of mechatronic systems relies on predictability, which is apparently at odds with the concept of emergent behavior. When designing mechatronic multi-agent systems, the main challenge thus lies in managing the inherent complexity and achieving predictability without ruling out the desired emergent behavior. We propose to achieve this by decomposing the requirements and design into largely independent aspects, realized as social structures with behavioral and communicative norms, and carefully composing them for each agent in such a way that the required analytic properties of one aspect cannot be invalidated by a second aspect. By explicitly grounding all abstract concepts in the concrete entities of an environment model of the mechatronic system, we can further support formal analysis and rapid prototyping.
Reference:
Advanced separation of concerns for mechatronic multi-agent systems through dynamic communities (Florian Klein, Holger Giese), In Proc. of the 3rd Workshop on Software Engineering for Large-Scale Multi-Agent Systems (in Conjunction with the International Conference on Software Engineering), Edinburgh, Scotland (Ricardo Choren, al, eds.), IEE, 2004.
Bibtex Entry:
@InProceedings{FKHG04b_ag,
AUTHOR = {Klein, Florian and Giese, Holger},
TITLE = {{Advanced separation of concerns for mechatronic multi-agent systems through dynamic communities}},
YEAR = {2004},
MONTH = {May},
BOOKTITLE = {Proc. of the 3rd Workshop on Software Engineering for Large-Scale Multi-Agent Systems (in Conjunction with the International Conference on Software Engineering), Edinburgh, Scotland},
PAGES = {112--119},
EDITOR = {Choren, Ricardo and al},
PUBLISHER = {IEE},
URL = {http://www.uni-paderborn.de/cs/ag-schaefer/Personen/Aktuell/Klein/Publications/2004/fkhg-SELMAS2004.pdf},
PDF = {uploads/pdf/1915-fkhg-selmas2004.pdf},
ABSTRACT = {Multi-agent systems present a promising paradigm for coping with the complexity of intelligent mechatronic applications. Frequently, they are characterized by the emergence of purposeful behavior and complex structures out of the seemingly chaotic interactions between their members. The safety of mechatronic systems relies on predictability, which is apparently at odds with the concept of emergent behavior. When designing mechatronic multi-agent systems, the main challenge thus lies in managing the inherent complexity and achieving predictability without ruling out the desired emergent behavior. We propose to achieve this by decomposing the requirements and design into largely independent aspects, realized as social structures with behavioral and communicative norms, and carefully composing them for each agent in such a way that the required analytic properties of one aspect cannot be invalidated by a second aspect. By explicitly grounding all abstract concepts in the concrete entities of an environment model of the mechatronic system, we can further support formal analysis and rapid prototyping.}
}