Dynamic Agent Architectures for Complex Systems

  • Authors:
  • Jeffrey Tweedale;Lakhmi Jain

  • Affiliations:
  • School of Electrical and Information Engineering, Knowledge Based Intelligent Engineering Systems Centre, University of South Australia, Mawson Lakes, SA 5095, Australia. {Jeffrey.Tweedale, Lakhmi ...;School of Electrical and Information Engineering, Knowledge Based Intelligent Engineering Systems Centre, University of South Australia, Mawson Lakes, SA 5095, Australia. {Jeffrey.Tweedale, Lakhmi ...

  • Venue:
  • Proceedings of the 2008 conference on Knowledge-Based Software Engineering: Proceedings of the Eighth Joint Conference on Knowledge-Based Software Engineering
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

Cognitive Science is a field of research attracting significant effort. This research requires the formation of teams of agents in order to dynamically configure the team with the ability to solve the decomposed task of the goal presented. Traditionally all tasks must be completed successfully or the team fails the goal [1,2]. A dynamic architecture would substitute agents within the team with alternative capabilities in order to succeed. It may even compromise and offer a partial solution and offer it to another system to complete. A good communications framework is required to pass messages between separate agent and other systems. Discussion about confined frameworks have recently been extended to enable individual students associated with our Knowledge-Based and Intelligent Information and Engineering Systems (KES) group to fast track the development of their research concepts. A Plug 'n' Play concept based on a multi-agent blackboard architecture forms the basis of this research. This paper highlights the core architecture, we believe is required for Multi-Agent System (MAS) developers achieve such flexibility. The research focuses on how agents can be teamed to provide the ability to adapt and dynamically organise the required functionality to automate in a team environment. The model is conceptual and is proposed initially as a blackboard model, where each element represents a block of functionality required to automate a process in order to complete a specific task. Discussion is limited to the formative work within the foundation layers of that framework.