ADAPTIVE CONTROL OF A ROBOTIC ARM

EXISTING ROBOTIC MOTION CONTROL TECHNOLOGY IS LIMITED IN ITS ABILITY TO HANDLE DYNAMIC EFFECTS (SUCH AS CHANGING PAYLOADS, INCREASED FRICTION DUE TO WEAR, ETC.) WITHOUT A DEGRADATION IN PERFORMANCE. IN ADDITION, THE TECHNIQUES CURRENTLY USED FOR REPOSITIONING ENDEFFECTORS IN SPACE CANNOT HANDLE CASES WHERE MULTIPLE POSSIBILITIES EXIST FOR THE INTERMEDIATE AND FINAL POSITIONS OF EACH JOINT OF THE ROBOT (I.E. THE EXISTENCE OF REDUNDANT DEGREES OF FREEDOM). HOWEVER, IT IS THROUGH THIS TYPE OF REDUNDANCY THAT FLEXIBILITY IS ACHIEVED TO ALLOW FOR SUCH THINGS AS COLLISION AVOIDANCE. THE OBJECTIVE OF THIS PROGRAM IS TO DEMONSTRATE A NOVEL APPROACH TO THE ADAPTIVE CONTROL OF A ROBOTIC ARM WITH REDUNDANT JOINTS. WE PROPOSE TO USE TECHNIQUES FROM THE FIELD OF ARTIFICIAL NEURAL SYSTEMS TO CONSTRUCT A SIMULATION OF THE ADAPTIVE CONTROL OF A ROBOTIC ARM. THE SIMULATION WOULD DEMONSTRATE THE ABILITY TO ADJUST TO CHANGES IN THE ROBOT'S PAYLOAD AND THE FRICTION IN ITS JOINTS, AS WELL AS THE ABILITY TO SMOOTHLY EXECUTE MOVEMENTS OF THE PAYLOAD THROUGH THE "BEST" SELECTION OF THE POSSIBLE JOINT VELOCITY PROFILES.