A Unified Approach to Planning Support in Hierarchical Coalitions

Clauirton de Albuquerque Siebra

(PhD Student)

Introduction | Planning Threads | Proposal | I-Kobe | Space Applications | Resources

Hierarchical coalitions are organisations whose components carry out different planning and plan execution activities at each level. One of the principal aims of knowledge-based tools for coalitions is to support such components so that they are able to work synergically together, each of them accounting for part of the planning and execution process.

The use of planning assistant agents is an appropriate option to provide this kind of support. Agents can extend the human abilities and be customised for different planning activities performed along hierarchical decision-making levels. However, the use of standard planning mechanisms is not sufficient to deal with the complexity of problems associated with coalition domains. In these domains, activities cannot consist merely of simultaneous and coordinated individual actions, but they must also be developed on a collaborative framework that ensures an effective mutual support among joint members.

This thesis analyses groups of requirements associated with the development of joint human-agent planning agents, showing that they can be implemented, in a unified way, via a constraint-based ontology and related functions. The constraints' properties have already been used by several planning approaches as an option to improve their efficiency and expressiveness. This work demonstrates that such properties can also be employed to implement collaborative concepts, which are kept transparent to the planning mechanisms. Furthermore, the use of constraints provides several facilities to the implementation of advanced mechanisms associated with the human interaction, as also demonstrated here.

The practical aspects of such an approach are illustrated via a prototype that uses a disaster relief domain as a test-bed. The role of this prototype is to show: (1) the impact of collaborative concepts in the planning process; (2) the opportunities for human-agent interaction, and; (3) the easy customisation of agents through the definition of activity handlers and specific constraint managers. Finally, an additional domain associated with space applications is also discussed so that we can testify the generality of this approach.

AIAI Artificial Intelligence Applications Institute
Centre for Intelligent Systems and their Applications
School of Informatics, The University of Edinburgh
Page maintained by c.siebra@ed.ac.uk, Last updated: Fri Sep 9 15:59:50 2005