Task skill formation via motion repetition in robotic dynamic manipulation

A system structure for acquiring the task skills in dynamic manipulation of objects using robotic manipulators is established, where the desired space trajectories do not need to be specified explicitly. A robotic batting is taken as a task example. The joint driving torque patterns of the manipulator are considered as the task skills and are learned against several typically given desired ball velocities. A multi-layered artificial neural network is used to learn and generalize the joint driving torque against various desired ball velocities, and an iterative optimal control algorithm is adapted to generate the supervisory joint driving torque signals for the neural network. Computer simulation and a three-degree-of-freedom manipulator is outlined and the results are depicted to explain the idea and verify the proposed approach