Nonlinear identification of a robotic arm using machine learning techniques

Abstract

With the advancement of intelligent algorithms more and more robots perform human tasks, be they due to dangerousness or simply by reducing human costs, for that to happen requires precision. This work has the objective of making an identification of a robotic arm with three phase induction motor through machine learning techniques to obtain a better model that represents the plant. The techniques used were Artificial Neural Network (ANNs): MLP, RBF and MLP + PSO. The techniques obtained a good performance, and they were evaluated through the multi-correlation coefficient (R2) for a comparative analysis.