NEURAL NETWORK CONTROLLER FOR ADAPTIVE MOVEMENTS IN ROBOTS

THE BIGGEST CHALLENGES IN CONTROLLING AUTONOMOUS ROBOTS TODAY DEAL WITH SELF-ORGANIZATION OF SENSORY-MOTOR COORDINATION, NOVELTY IN THE WORKING ENVIRONMENT, AND PROCESSOR FAULTS. TO MEET THESE NEEDS, THE PROPOSED STUDY WILL MODEL A PROTOTYPE NEURAL ARCHITECTURE USED TO CONTROL THE DYNAMIC COORDINATION OF A MULTIJOINT ROBOT ARM AND TWO STEREO CAMERAS. RESEARCH ON THE PROTOTYPE DYNAMIC CONTROLLER WILL BE BASED ON A WORKING VERSION OF A MODEL CONTROLLER THAT CAN ALREADY SELF-ORGANIZE ARM POSTURES GUIDED BY TWO CAMERAS. THE OBJECTIVE OF THE PROPOSED STUDY IS TO EXTEND THE PREVIOUS STATIC MODEL INTO A DYNAMIC MODEL, WHICH WILL BE ABLE TO GENERATE ADAPTIVE TRAJECTORIES. WHEN IMPLEMENTED, THIS PROTOTYPE SYSTEM WILL COORDINATE A MULTI-JOINT ROBOT ARM TO ADAPTIVELY REACH TARGETS IN THREE DIMENSIONS IN REAL TIME. THE SYSTEM WILL SELF-ORGANIZE AND MAINTAIN VISUAL-MOTOR CALIBRATIONS STARTING WITH ONLY LOOSELY DEFINED RELATIONSHIPS. THE SYSTEM WILL TOLERATE INTERNAL NOISE, PARTIAL SYSTEM DAMAGE AND CHANGES IN THE MECHANICAL AND OPTICAL PARAMETERS OF THE ROBOT AS THEY OCCURDURING WEAR. THIS ADAPTABILITY REPLACES THE NEED FOR OPERATOR CALIBRATION. IN PHASE II OF THIS PROJECT, THIS PROTOTYPE ROBOT CONTROLLERWILL BE BUILT AND TESTED.