Distributed Case-Based Planning: Multi-Agent Cooperation for High Autonomy

J. Britanik and M. Marefat

Department of Electrical and Computer Engineering
University of Arizona
Tucson, Arizona 85721

Email: { britanik,marefat} @ece.arizona.edu


A drawback of typical knowledge-based systems, including case-based reasoning systems, is that human intervention is often unavoidable when there is a lack of knowledge necessary to solve a problem. Our approach described here introduces the concept of cooperative case-based reasoning, where a case-based system may interact with another case-based system or other knowledge based system to acquire the missing knowledge it needs to solve a problem.

The basic mechanism by which this knowledge transfer takes place is the knowledge interface. The knowledge interface translates information in a knowledge-based system and places it into a globally recognized format; likewise, the knowledge interface converts information in the globally recognized format and places it into the context of the local system.