Improving efficiency of the Newton algorithm of motion planning for robot manipulators

In this paper a method to speed up a convergence of the Newton algorithm of motion planning for manipulators was presented. The method couples one dimensional optimization (with respect to a coefficient influencing the convergence property of the algorithm) with a virtual goal replacing a real goal of the planning. The first modification of the basic Newton algorithm can be used for any taskspace while the second one works only for low dimensional (not greater than 3) taskspaces. An algorithm based on the method was provided and its efficiency was illustrated on a planar double pendulum manipulator.