NEURAL NETWORK BASED TARGET RECOGNITION

REAL-TIME OBJECT ACQUISITION, TRACKING AND RECOGNITION REQUIRE MASSIVE COMPUTATIONAL BANDWIDTH. WE PROPOSE TO RESEARCH AND DEVELOP CUSTOM INTEGRATED CIRCUITS THAT WILL PERFORM THESE TASKS AT THE FOCAL PLANE, USING INTEGRATED PHOTOSENSORS AND ANALOG COMPUTING NETWORKS. PHASE I WILL INCLUDE THE DESIGN AND FABRICATION OF TET CHIPS TO PROVE THE FEASIBILITY OF OUR DEVELOPMENT. OUR APPROACH WILL UTILIZE STANDARD READILY AVAILABLE CMOS BULK INTEGRATED CIRCUIT TECHNOLOGY SO PRODUCTS ARISING FROM THIS R&D CAN BE FABRICATED RELIABLY AND ECONOMICALLY BY A NUMBER OF VENDORS. WE HAVE ALREADY DEMONSTRATED USING THIS TECHNOLOGY TO FABRICATE CONTINUOUS-TIME ANALOG COMPUTING CIRCUITS INTEGRATED WITH PHOTOSENSORS TO EXTRACT VELOCITY OF A MOVING IMAGE. WE HAVE ALSO SHOWN THE FEASIBILITY OF USING FLOADING-NODES TO PROVIDE NON-VOLATILE STORAGE OF ANALOG VOLTAGES. THESE VOLTAGES CAN BE USED TO CONTROL THE OPERATION OF THE CHIP THUS IMPLEMENTING THE CAPABILITY TO LEARN OR BE TRAINED IN THE FIELD. THE SAME ADAPTATION MECHANISM CAN BE USED TO COMPENSATE FOR FABRICATION NON-UNIFORMITIES. THIS PROPOSED RESEARCH APPLIES OUR EXPERIENCE TO THE CHALLENGING AND IMPORTANT PROBLEM OF REAL-TIME RECOGNITION AND MAY LEAD DIRECTLY TO INEXPENSIVE PRODUCTS WITH WIDE APPLICABILITY COMMERCIALLY AND TO THE MILITARY.