Inverse Kinematics Using Particle Swarm Optimization, A Statistical Analysis

Abstract

This paper investigates the relative performances of PSO variants when used to solve inverse kinematics. Inverse kinematics is a key issue in robotics; for problems such as path planning, motion generation or trajectories optimization, they are classically involved. In the specific case of articulated robotics, inverse kinematics is needed to generate the joint motions, correspondent to a known target position. Articulated systems are very important in humanoid robotics, since arms and legs belong typically to this kind of mechanisms. In this paper the IK-PSO, Inverse Kinematics PSO, is applied to a double link articulated system. A statistical analysis is conducted to survey the convergence and relative performances of the main PSO variants when applied to solve IK; the PO variants tested are: Inertia weight PSO, Constriction factor PSO, linear decreasing weight and two simplified PSO variants.