Inverse kinematics using single- and multi-objective particle swarm optimization

Robotic manipulators are currently extensively used in a wide spectrum of industrial, medical and commercial facilities. While it is becoming a trend to use over-actuated manipulators, solution of related inverse kinematics problems becomes more challenging. Recently, more efforts have been directed towards utilizing alternative computationally efficient methods for solving such problems. During the last decade, the use of evolutionary algorithms involving particle swarm optimization as well as neural networks has been on the rise as a plausible alternative. This paper presents a new choice of parameters for the single-objective particle swarm optimization that leads to accurate and computationally efficient solutions. Furthermore, a multi-objective particle swarm optimization technique, which offers more flexibility in the manipulator motion, is introduced. Mathematical formulation of the problem and applications on different robotic manipulators are presented in the paper with good simulation results.