ARTIFICIAL INTELLIGENCE (AI) DEVELOPMENT FOR PILOT AID APPLICATIONS

THE SUCCESSFUL INTRODUCTION OF AI TECHNOLOGY INTO AIR FORCE AVIONICS HAS BEEN HINDERED BY THE NEED FOR REAL-TIME REASONING WITH INCOMPLETE AND INCONSISTENT DATA, INTEGRATION OF HIGH LEVEL SUBSYSTEMS WITH SYSTEMS FOR SIGNAL ANALYSIS, AND AN ADAPTIVE HUMAN-COMPUTER INTERFACE. DSC PROPOSES TO TEST THE HYPOTHESIS THAT SOLUTIONS FOR ALL THREE PROBLEMS ARE LINKED, AND THAT IMPROVED ARCHITECTURES CAN BE DEVELOPED BY INCORPORATING FEATURES FROM FUZZY LOGIC, BELIEF FUNCTIONS, RECENT WORK IN BAYESIAN PROBABILITY, AND NON-MONOTONIC REASONING. THE PRIMARY OBJECTIVE OF PHASE I RESEARCH IS TO EXPLORE THE FEASIBILITY OF DEVELOPING NEW METHODS FOR INFERENCE IN AVIONICS EXPERT SYSTEMS, WHICH ADDRESS SPECIFIC SHORTCOMINGS IN EXISTING APPROACHES AND COMBINE SOME OF THEIR DISTINCT VIRTUES. THE SECOND OBJECTIVE OF PHASE I RESEARCH IS TO IMPROVE UNDERSTANDING OF THE CHARACTERISTICS OF AVIONICS APPLICATION DOMAINS THAT MAKE ONE INFERENCE SCHEME MORE APPROPRIATE THAN ANOTHER. FIVE TASKS HAVE BEEN PROPOSED: (1) A SYSTEMATIC EXAMINATION OF ALTERNATIVE INFERENCE MECHANISMS IN TERMS OF FACTORS RELEVANT TO AVIONICS PILOT AID APPLICATIONS; (2) DEVELOPMENT OF IMPROVED PILOT AID INFERENCE FRAMEWORK; (3) DEVELOPMENT OF CONCEPTS FOR ADAPTIVE HUMAN-COMPUTER INTERACTIONS; (4) IMPLEMENTATION OF A SMALL-SCALE PROTOTYPE SYSTEM IN AN OPERATIONAL CONTEXT; (5) DEMONSTRATION AND TESTING OF ALTERNATIVE INFERENCE SCHEMES IN THE SELECTED CONTEXT.