RBF Neural Network Based Kinematic Calibration of a Planar Parallel Robot

Abstract

This paper presents the kinematic calibration of a planar parallel robot. A radial based function (RBF) neural network based nonparametric method is proposed, in which the network is used to store and interpolate the joint correction. The experimental results show that it works more effectively than nonlinear regression based model parameter identification and spline interpolation based joint correction. This is because the method is free from validity of model and approximates the kinematic behavior of the actual robot more accurately. The accuracy is improved from 1.66 mm (maximum) and 0.99 (average) mm to 0.0284 (maximum) 0.0158 mm (average) by the proposed method