Battle Management Based On Autonomous Agents

BATTLE MANAGEMENT CAN BE VIEWED, AT LEAST IN PART, AS A COMPLEX OPTIMIZATION PROBLEM WHERE THERE ARE THREATS OF VARYING CAPABILITY, AND THERE ARE ASSETS WHICH CAN BE USED TO SENSE AND ENGAGE OR INTERCEPT THE THREATS. THE PROBLEM IS COMPLICATED BY THE FACT THAT MOST OF THE SENSOR INFORMATION HAS SIGNIFICANCE UNCERTAINTY AND NOISE DUE TO THE PHYSICAL LIMITATIONS OF THE SENSORS, COUNTERMEASURES, AND IN SOME BATTLE SCENARIOS, NUCLEAR EFFECTS. CONVENTIONAL OPTIMIXATION TECHNIQUES ARE TOO SLOW FOR THE COMMAND CENTER NEEDS, DO NOT YIELD GOOD RESULTS IN THE PRESENCE OF UNCERTAIN AND NOISY DATA, AND ARE BRITTLE IN THE PRESENCE OF UNANTICIPATED SITUATIONS. SDIO BATTLE MANAGEMENT THEN BECOMES ONE OF NOT A MATHEMATICAL OPTIMIZATION, BUT ONE OF HEURISTIC REASONING, REASONING UNDER UNCERTAINTY, AND "RULES OF THUMB." THE PROBLEM DOMAIN APPROACHES THAT OF "DETERMINISTIC CHAOS" WHERE A VERY SMALL CHANGE IN SOME PARAMETER MAY MAKE A MUCH LARGER CHANGE IN THE OUTCOME. THIS PROPOSAL DETAILS HOW WE WILL USE "AUTONOMOUS AGENTS" TO MODEL THE VARIOUS ELEMENTS WITHIN A BATTLE SITUATION, AND THEN USE THESE AUTONOMOUS AGEMTS TO EITHER GENERATE NEAR OPTIMAL BATTLE STRATEGY, OR PROVIDE DECISION AIDS TO AN OPERATOR. THE PROPOSAL ATTEMPTS TO DEMONSTRATE THAT THIS APPROACH WILL PROVIDE ROBUST, NEAR OPTIMAL RESULTS EVEN IN THE CASE THAT THE SYSTEM IS DETERMINISTICALLY CHAOTIC. THE "AUTONOMOUS AGENT" PARADIGM HEREIN PROPOSED FOR BATTLE MANAGEMENT IS VERY MUCH AN AI TECHNIQUE. THE LITERATURE CONTAINS EXCELLENT EXAMPLES OF HOW AUTONOMOUS AGENTS HAVE BEEN USED TO REPRODUCE SELECTED INTELLIGENT BEHAVIOR OF BIOLOGICAL ORGANISMS. OUR PROPOSAL DETAILS HOW THIS SAME PARADIGM WILL BE USED FOR BATTLE MANAGEMENT APPLICATIONS.