KNOWLEDGE NETWORKS FOR MISSION PLANNING AND FLIGHT CONTROL

MISSION PLANNING AND FLIGHT CONTROL INVOLVE THE COORDINATION OF BOTH CONCEPTUAL AND PHYSICAL VARIABLES. THE FUZZY COGNITIVE MAP (FCM) FRAMEWORK IS PROPOSED AS A FLEXIBLE ARCHITECTURE FOR COORDINATING THESE VARIABLES IN REALTIME PARALLEL SYSTEMS. FCMS ARE ADAPTIVE KNOWLEDGE NETWORKS FOR REPRESENTING UNCERTAIN GOAL, POLICY, UTILITY, CAUSAL, RESOURCE, AND PHYSICAL VARIABLES THAT INTERACT IN THE MISSION PLANNING/FLIGHT CONTROL PROCESS. FCMS SYNTHESIZE EXPERT-SYSTEM SEARCH TREE AND NEURAL NETWORK TECHNOLOGIES IN A NOVEL, EXPRESSIVE FRAMEWORK. FCM KNOWLEDGE ACQUISITION REDUCES TO CAUSAL-POLICY PICTURE DRAWING AND IMPROVES, NOT DEGRADES, WITH INCREASING NUMBERS OF EXPERTS. FCM INFERENTIAL DYNAMICS RESEMBLE NEURAL NETWORK COMPUTATIONS. THE FCM MATRIX STRUCTURE ALLOWS EACH MATRIX COLUMN TO BE CONCURRENTLY PROCESSED. NEURAL NETWORKSCAN ESTIMATE FCM VARIABLES, SUCH AS MISSION SUCCESS RATES, FROM HISTORICAL AND ON-LINE DATA. EXPERT SYSTEM META-RULES CAN GOVERN THE FIRING OF DIFFERENT FCMS, AND VICE VERSA, IN COMPLEX HYBRID SYSTEMS. FUZZY CONTROL SUBSYSTEMS CAN BE ACTIVATED IN PARALLEL BY FCMS AND CAN IN TURN ACTIVATE EMBEDDED FCMS TO DIFFERENT DEGREES.