Accelerating Newton Algorithms of Inverse Kinematics for Robot Manipulators

In this paper a few methods of accelerating computations of an inverse kinematic task for robot manipulators are discussed and compared. The methods adapt known ideas to the domain of robotics, and rely on an approximation of inverse matrices, that are computed in each iteration of the Newton algorithm of inverse kinematics, with much computationally cheaper matrix polynomials. The computational efficiency of the methods is illustrated on planar pendula with a varied degree of redundancy.