Numerical Algorithm for Solving Robot Inverse Kinematics

摘要

An extension of the inverse kinematics algorithm by Gupta and Kazerounian (1985) is presented. The robot kinematics is formulated by using the Zero Reference Position Method (Gupta, 1981). Euler parameters and the related vector forms of the spatial rotation formula and rotation concatenation have been used to improve the efficiency of the velocity Jacobian computation. The joint rates are formally integrated by using a modified predictor-corrector method particularized to robot inverse kinematics--it is a strictly descending, p(1) c (o-n), variable step algorithm. The definition of the deviational error measure has been revised. Depending upon the convergence criteria used, these modifications reduce overall computational times by 15-25%.