Adaptive manipulator control (movement learning algorithms)

Iterative procedures are proposed for piecewise continuous control (learning procedure) which moves the manipulator grip from the initial position into a final one. The algorithms can be implemented in the manipulator itself and do not require preliminary determination of the system parameters (dimensions, masses, momenta, viscosity of the medium, etc.). The algorithm is proved to converge and the rate of convergence is determined. A method for increasing the response of the resultant control is described.