INTELLIGENT TASK PLANNING AND EXECUTION FOR ROBOTIC CONTROL

THE OBJECTIVE OF THIS RESEARCH IS THE CONCEPTUAL DESIGN OF CAPABILITIES FOR TASK PLANNING AND PLAN EXECUTION IN A HIERARCHIAL ROBOTIC CONTROL SYSTEM. THE ROBOTIC CONTROL SYSTEM HAS MULTISENSORY CAPABILITIES INCLUDING FORCE/TORQUE, TACTILE AND VISION. LEVERAGE IS GAINED IN PLANNING EFFICIENCY BY STRUCTURING KNOWLEDGE OF 1 PREVIOUSLY USEFUL NAMED PLANS, PLANBOXES AND SCRIPTS TO REDUCE THE EFFORT NEEDED FOR REPLANNING. THE PLANS GENERATED MAY ALSO BE RECALLED OR LATER USE AS SCRIPTS. ADVANTAGES ARE ALSO TO BE GAINED IN ADAPTABILITY BY PUTTING OFF DECISIONS UNTIL EXECUTION TIME. THIS IS BECAUSE FEEDBACK FROM THE ENVIRONMENT CAN CONSTRAIN THE PLAN AND OFTEN OBVIATE THE NEED FOR REPLANNING WHEN SITUATIONS IN THE ENVIRONMENT ARE DIFFERENT THAN EXPECTED.