Modeling Time-Stressed Situation Assessment and Planning for Intelligent Opponent Simulation

摘要

This report summarizes research undertaken to simulate expertise on a real-time time-stressed task requiring situation assessment and planning. While prior efforts in expert systems development have attacked a variety of problems either in situation interpretation or planning, none has attempted to articulate a model that can support both types of activities simultaneously in a dynamic environment. The present study attempted to derive such a cognitive model for time-stressed performance in the domain of Naval surface-warfare tactical decision making. The model extends the concept of the 'blackboard' architecture--multiple independent processing specialists contributing data and results to a common knowledge base--previously applied to systems for either signal interpretation or planning. Our approach treats the time-stressed assessment and planning task as a hierarchy of blackboard models, with several processing modules--each organized as a blackboard model--interacting through a common knowledge base. Implementation of the model is underway for use as an automated opponent in a Navy tactical battle game. The simulation will emulate an opposing naval commander and will be embedded in an interactive instructional system providing battle exercises for Navy tactics trainees. Author keywords include: Intelligent Computer-Assisted Instruction (ICAI), artificial intelligence, rule-based systems, knowledge sources, scheduling, wargaming, and LISP.