Combining a Differential Evolution Algorithm with Cyclic Coordinate Descent for Inverse Kinematics of Manipulator Robot

Abstract

This paper presents a new approach to solving inverse kinematics of seven degree-of-freedom (DOF) robotic manipulators. The inverse kinematic (IK) of a manipulator robot determines the joint values for which the end-effector reaches a given position and orientation. The manipulator with seven or more degrees of freedom have infinitely many solutions and more difficult to find the inverse kinematics. We propose a combination of a new compact differential evolution algorithm with the cyclic coordinate descent (CCD) method to solve the IK problem. The CCD is an iterative method that optimize for one joint variable at a time. A differential evolution algorithm is a meta-heuristic optimization. The combination of the CCD and a new differential evolution algorithm is performed in two steps. The proposed method is not sensitive to singularity configurations, and not subject to numerical instabilities. The combined method can converge to the approximately precise position with virtually any initial positions. To evaluate the performance of the proposed algorithm, computer modeling and simulations of a seven DOF manipulator were carried out by solving inverse kinematic problems with various target positions. From the experimental results, the proposed methods deliver more precise solutions than the general pseudo inverse Jacobian that currently in use.