A memetic differential evolution algorithm for the inverse kinematics problem of robot manipulators

Abstract

The inverse kinematics (IK) of robot manipulators defines the problem of finding the joint configuration that places the end-effector in an arbitrary position and orientation. This problem can be formulated in the configuration space as a constrained optimisation model to avoid the numerical instability of Jacobian-based IK methods. Previous works have shown that differential evolution (DE) is effective in finding accurate solutions to the IK optimisation problem, but exhibits low convergence speed rates. In this paper we propose a memetic differential evolution (dDE) to improve the convergence behaviour of the standard DE scheme. Both algorithms are tested and compared in a simulation environment as kinematic inversion methods for two non-redundant robot manipulators. Results revealed that dDE outperforms the original DE algorithm in accuracy and convergence speed.