Issues and Approaches to Planning under Uncertainty

摘要

This project has addressed the use of artificial intelligence (AI) techniques as the foundation for both situation assessment in naval-task-force engagements, and for planning effective combat operations in such engagements. Developing an effective automated planning agent requires progress in a number of areas, including the integration of data received from a wide variety of sensor types dispersed over large geographical areas with other sources of knowledge, and the planning, execution, and monitoring of plans involving dispersed actors with a variety of capabilities. The difficulties associated with these tasks are exacerbated by the fact that the system must operate effectively in situations where the degree of intercommunication between the participants may be severely constrained, implying an inability to rely on centralized command and control. Our work has addressed the following points: Developing a knowledge base of anticipated actions. Characterizing the gross properties of shipborne/airborne sensors. Investigating alternative approaches to planning -- particularly, hierarchical techniques. Developing an inference capability for linking disparate knowledge sources.