Abstract
Shifting matrix management (SMM) is a model of agent coordination inspired by Mintzberg's model of organizational structures. Mintzberg's model permits many temporary lines of authority, reflecting the multiple and shifting functions of a flexible workforce. In order to apply these ideas to agent cooperation, a six-stage framework has been devised. The resulting model has been compared with two standard models: Contract Nets and Cooperative Problem-Solving. All three models have been implemented by means of an in-house blackboard system, Algorithmic and Rule-based Blackboard System (ARBS). 'Disembodied' agents have been constructed whose components are spread between system modules, known as knowledge sources, and private partitions of the blackboard. Tests have been carried out in which the three models have been applied to a set of tasks involving the control of two robots. Within the narrow context of these tests, the SMM model out-performs the other two approaches in terms of its task completion rate, number of tasks completed, and avoidance of wasted efforts. It is argued that although the SMM model expends more time reasoning about its actions, this is likely to be more than offset by the resultant efficient use of resources.
Original language | English |
---|---|
Pages (from-to) | 191-201 |
Number of pages | 11 |
Journal | Engineering Applications of Artificial Intelligence |
Volume | 16 |
Issue number | 3 |
DOIs | |
Publication status | Published - Apr 2003 |
Externally published | Yes |
Keywords
- Agent
- ARBS
- Blackboard system
- Contract Nets
- Cooperative problem-solving (CPS)
- Disembodied agents
- Distributed artificial intelligence
- Shifting matrix management (SMM)