A closed-loop neuro-parametric methodology for the calibration of a 5 DOF measuring robot

Abstract

The paper deals with the calibration of industrial robots, a very important issue in robotics. A methodology to improve accuracy obtained by classical parametric methodologies is proposed. The method is based on the application of a neural network together with a classical parametric model of the robot kinematic. Due to this combination of methodologies the approach could be defined as "hybrid neuro-parametric method". Experimental results prove an improvement in the robot accuracy.