Application of ensemble learning approach in function approximation for dimensional synthesis of a 6 DOF parallel manipulator

Abstract

Presented in this paper is an investigation of the use of ensemble methods in machine learning for developing function approximation models of the analytical objective function, to be applied to an optimization search process of a 6 DOF parallel manipulator. The process of optimization of these mechanisms can be cumbersome, as it often involves complex objective functions and diverse design parameters. The use of ensemble methods in machine learning methods combination is demonstrated and evaluated against the individual or base methods using dataset from a parallel robotic manipulator. Experiments are carried out to determine whether an ensemble performs better than the base methods.