Our framework for directability enables the declaration of preferences and restrictions on agent activity. Our focus is on three main areas: (a) strategy preferences that describe the approaches to be used in executing assigned tasks; (b) adjustable levels of agent autonomy; and (c) reactive responses in the face of unexpected events. Successful delegation further requires human visibility into agent behavior. For this reason, we are developing techniques for supporting variable reporting by agents. These techniques enable agents to adjust the frequency and detail of information that they transmit regarding the status of their assigned tasks.
Publications
Policy-based Agent Directability, K. L. Myers and David N. Morley.
To appear in Adjustable Autonomy, edited by H. Hexmoor, R. Falcone,
and C. Castelfranchi, Kluwer Academic Publishers, 2002.
[PDF]
Resolving Conflicts in Agent Guidance, K. L. Myers and David N. Morley. Proceedings of the AAAI-02 Workshop Preferences in AI and CP: Symbolic Approaches, 2002. [PostScript] [PDF]
Conflict Management for Agent Guidance, K. L. Myers and David N. Morley. Proceedings of the First International Joint Conference on Autonomous Agents and Multi-Agent Systems (AAMAS), 2002. [PostScript] [PDF]
Human Directability of Agents, K. L. Myers and David N. Morley. Proceedings of the First International Conference on Knowledge Capture, K-CAP 2001, Victoria, B.C., Canada, 2001. [PostScript] [PDF ]
Directing Agent Communities: An Initial Framework, K. L. Myers and David N. Morley. Proceedings of the IJCAI-01 Workshop Autonomy, Delegation and Control: Interacting with Autonomous Agents, 2001. [PostScript] [PDF]
User's Guide for the TIGER System, David N Morley, April 2002. [ Postscript ] [ PDF ]
Acknowledgments
This project has been funded by DARPA's Control of Agent-based Systems
program through contract F30602-98-C-0160 under the supervision of Air
Force Research Lab - Rome.
Last modified: December, 2000