Industrial manipulator dynamic parameter estimation using mutating particle swarm optimization (Mupso)

Abstract

This work aims at developing a dynamic model and estimating the unknown parameters of the first three joints (at the arm) of a 6 degree of freedom industrial robot manipulator, a finite Fourier series algorithm was used to design an excitation trajectory, a mutating particle swarm optimization algorithm was used to optimise the parameters of the Fourier series thereby minimizing the condition number of the observation matrix, and a linear least-squares methods was implemented for estimating the unknown dynamic parameters of the manipulator. A mutation function was implemented to break the algorithm out of stagnation. Out of the thirty unknown parameters at the industrial manipulator arm, twenty were identified independently, two were identifiable in linear combinations, and the remaining eight parameters were unidentifiable. The mutating particle swarm optimization algorithm dominated other algorithms and was found suitable for robot dynamic analysis.