ROBOT VISION SYSTEMS BASED ON LOG-POLAR IMAGE PLANE COORDINATES

LOG-POLAR COORDINATES HAVE RECENTLY BEEN SHOWN TO POSSESS SIGNIFICANT COMPUTATIONAL ADVANTAGES OVER TRADITIONAL CARTESIAN COORDINATES IN EXPRESSING TRANSFORMATIONS ASSOCIATED WITH DYNAMIC 3-D PERSPECTIVE IMAGERY. REDUCED PIXEL COUNT AND REDUCTION IN CALCULATIONS PER PIXEL SUGGEST THE FEASIBILITY OF COMPACT, HIGH-PERFORMANCE SYSTEMS NOT POSSIBLE WITH CARTESIAN SYSTEMS. RECENT THEORETICAL DEVELOPMENTS INCLUDE THE FORMULATION OF NEW ALGORITHMS FOR VISUAL TRACKING AND DEPTH MAPPING. RECENT HARDWARE DEVELOPMENT INCLUDE THE CONSTRUCTION OF VIDEO RATE REMAPPERS WHICH CONVERT CARTESIAN IMAGERY TO LOG-POLAR COORDINATES AND THE WORLD'S FIRST CCD CHIP TESSELLATED IN LOG-POLAR COORDINATES. THE TIME HAS COME TO INTEGRATE THESE DEVELOPMENTS IN THE DESIGN OF A NEW TYPE OF ROBOT VISION SYSTEM. PHASE I WILL DESIGN A ROBOT VISION TESTBED WHICH INTEGRATES THE NEW ALGORITHMS IN THE PERFORMANCE OF USEFUL VISUAL FUNCTIONS SUCH AS DYNAMIC DEPTH MAPPING AND CAMERA/VEHICLE MOTION CONTROL. PHASE I METHODS INCLUDE MATHEMATICS, SIMULATION AND SYSTEM DESIGN. PHASE II INCLUDES CONSTRUCTION OF A PROTOTYPE SYSTEM WHICH ROBUSTLY AND EFFICIENTLY PERFORMS SUCH FUNCTIONS. ULTIMATE BENEFITS INCLUDE ROBOT VISION SYSTEMS THREE ORDERS OF MAGNITUDE MORE COMPACT THAN THOSE POSSIBLE USING CARTESIAN COORDINATES, OPENING THE DOOR TO COST-EFFECTIVE COMMERCIAL ROBOTS WITH POWERFUL VISION SYSTEMS OCCUPYING ONLY A FEW CUBIC INCHES.