A First-Order Computational Model of Human Operational Planning
Abstract
Most computational models of planning developed within Al require complete and accurate knowledge of the starting situation as well as a well-defined goal for their operation. In marked contrast. human planners often must create plans despite the fact that the goal is often ill-defined and knowledge about the world in which the plan is to be executed is incomplete and possibly inaccurate. Battlefield planning is certainly a paradigmatic example of such a planning context. Consequently, the development of a model of how planning takes place within this type of domain represents a considerable challenge to our current computational understanding of planning. The research reported here is directed at this challenge. Based on observations made of the planning activities of human experts at the U.S. Army War College and a review of the pedagogy associated with the training of human experts, the present paper provides a basic first-order model of the possible control strategy employed by experts in battlefield planning. Next, the purpose model is evaluated by comparing its "predictions" concerning the types of planning activities that should be observed and the order in which these activities should appear against a protocol of the planning activities of human experts. Implications of this first-order model for further knowledge acquisition and development of planning aids are discussed.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1988
- Accession Number
- ADA395978
Entities
People
- Charles F. Schmidt
- Gerald M. Powell
Organizations
- United States Army Communications-Electronics Command