The Nicole Project: Agent Theory Compendium
The following is an attempt to sum up a great deal of theories about agents and agent architectures. Currently, its structure
is based on the MBOR theory, but the outline and information should be expanded.
*THIS DOCUMENT IS EXTREMELY PRELIMINARY AND IS PRIMARILY A PLACEHOLDER*.
For a more complete and up-to-date discussion of these issues, PLEASE consult
the
Memory-Based Opportunistic Reasoning Home Page
or its postscript form at:
The Characteristics of the World
- regular
- uncertain
- dynamic
- "action-feedback loops"
- relevant to agent goals
- "ill-structured problems and situations"
- "shifting, ill-defined or competing goals"
- resource-bounded
- ?
Desired Properties of Agents
- sensation
- action
- memory/learning
- determine relevance of environment to agent goals
- situate action to context
- reason about goals, situations, actions, desires
- ?
A Theory of Agent Architecture
- Memory
- persistent
- accretive/experiential
- expressive
- context-sensitive
- resource-bounded
- asynchronous
- ?
- Reasoning
- agent ontology
- knowlege goal generation
- integration mechanisms
- ?
- Control
- specification of agent capabilities
- specification of utility of actions
- map intentions to capabilities and utility values
- updatable decision procedure
- context-sensitive decision procedure
- ?
What is an Agent?
- c.f. agent definition appendix
Direct academic correspondence to:
Anthony G. Francis, Jr.
AI / Cognitive-Science Group
College of Computing
Georgia Institute of Technology
Atlanta, Georgia 30332-0280
Phone: (404) 853-9381, (404) 853-9372 fax
E-mail: centaur@cc.gatech.edu
For more information, try `finger'
or contact me directly via e-mail to centaur@cc.gatech.edu.
Last Modified: August 10, 1995 by Anthony Francis (centaur@cc.gatech.edu)