Neural Network Solutions for Forward Kinematics Analysis of 2-(6UPS) Manipulator

In this paper, we present Kinematic analysis of a specific class of series-parallel manipulators, known as 2(6-UPS) manipulator, which composed of two modules which consist of elementary manipulators with the parallel structure of the Stewart Platform. After extracting the Kinematics Model of the hybrid manipulator, because of highly nonlinear relations between joint variables and position and orientation (location) of the end effectors in this kind of hybrid mechanism, we applied wavelet neural network with its inherent learning ability, to solve the kinematics problem. Proposed wavelet neural network (WNN) is applied to approximate the paths of mid and upper plate in different paths. Comparison the results of neural network with closed form solution (CFS) show high accurate performance of proposed WNN.