SYSTEM LOGISTIC PLANNING AND DECISION AID FOR THE BATTLEFIELD COMMANDER

COMMAND IN GROUNDD COMBAT INVOLVES MANY DIFFICULT DECISION-MAKING TASKS. A HIGHLY PROMISING ROUTE FOR ASSISTANCE TOTHE COMMANDER IS USEOF COMPUTERS. BY 1985-87 GREAT COMPUTATIONAL POWER WILL BE AVAILABLE IN LAP-SIZED MACHINES; HOWEVER, SUBSTANTIAL ADVANCES IN SOFTWARE ARE NECESSARY IF THIS PROMISE IS TO BE REALIZED. IN THIS PROPOSAL, DEFENSE SYSTEMS, INC. (DSI) DESCRIBES RESEARCH WHICHWILL LEAD TO DEVELOPMENT OF AN OPERATIONALLY USEFUL AID WHILE ADVANCING THE STATE OF THE ART IN AID DESIGN. THIS EFFORT WILL EXPLOIT THE POTENTIAL OF KNOWLEDGE-BASED ARTIFICIAL INTELLIGENCE (AI) SYSTEMS TO BUILD A FLEXIBLE, SELF-ADAPTIVE PLANNING DEVICE. BECAUSE AI RESEARCH IS MOST SUCCESSFUL IN TIGHTLY BOUND, FORMALLY WELL-UNDERSTOOD AREAS, DSI WILL FIRST DEVELOP A LOGISTICS PLANNING AID. TECHNIQUES ESTA BLISHED IN THIS EFFORT CAN THEN BE APPLIED TO LESS WELL-STRUCTURED DOMAINS, SUCH AS OPERATIONS. THE AID WILL CONSIST OF A "SMART" FRONT END AND A LIBRARY OF LOGISTICS SUBROUTINES. THE TECHNICALLY INTERESTING PORTION OF THE WORK WILL INVOLVE THE DEVELOPMENT OF THE FRONT END, WHICH WILL: (1) HELP A NAIVE USER SET UP LOGISTICS PROBLEMS; (2) AUTONOMOUSLY SELECT AND EMPLOY ALGORITHMS BEST SUITED TO SOLVE THEM; AND (3) LEARN ABOUT ITS ENVIRONMENT DURING USE. THE CHIEF METHOD BY WHICH THE AID WILL LEARN IS THROUGH USE OF AN ADAPTIVE LEARNING NETWORK APPROACH TO HELP THE USER DECIDE WHEN TO REFORECAST DEMAND FOR SUPPLIES.