PARALLEL IMPLEMENTATION OF ALGORITHMS FOR ROBOTIC SENSORY FUSION

THE FOCUS OF THIS PROPOSAL IS THE PARALLEL IMPLEMENTATION OF ALGORITHMS FOR ROBOTIC SENSORY FUSION. WE PROPOSE TO INVESTIGATE THE PARALLEL IMPLEMENTATION AND BENCHMARKING OF MULTISENSOR AND PARALLEL ROBOTIC/AUTONOMOUS TASKS INVOLVING THE USE OF SENSORS SUCH AS VIDEO, RANGE, PROXIMITY, TOUCH, FORCE, AND TORQUE. THE INNOVATION IN THIS PROJECT IS THE PARALLEL MULTISENSING AS WELL AS THE PARALLEL AND PROCESSINGOF SENSORY CUES TO ALLOW FAST IDENTIFICATION OF OBJECTS IN VARIABLY STRUCTURED OR UNSTRUCTURED WORKSPACES TO PERFORM REALISTIC ROBOTIC TASKS THAT REQUIRE MULTISENSOR FUSION. THE NEED FOR THIS SENSORY AND COMPUTER TECHNOLOGY CUTS ACROSS THE BOARD OF NASA ACTIVITIES INCLUDING DEPLOYMENT, MAINTENANCE, AND REPAIR MISSIONS. THE EQUIPMENT NECESSARY FOR THE EXECUTION OF THIS PROJECT IS IN PLACE. PERCEPTICS HAS RECENTLY INTEGRATED ONE OF THE LARGEST MASSIVELY PARALLEL COMPUTERS IN THE NATION: A 1024-PROCESSOR NCUBE HYPERCUBE. THE UNIVERSITY OF TENNESSEE HAS A VERSATILE AND COMPLETE ROBOTIC WORKSTATION EQUIPPED WITH ALL THE SENSORS NECESSARY FOR THIS PROJECT. AS A RESULT, THE CHANCES OF TECHNICAL SUCCESS AND LIKELIHOOD OF COMMERCIALIZATION OF THE OUTCOME OF THIS PROJECT ARE HIGH.