Neural Network Solutions for Forward Kinematics Problem of Hybrid Serial-Parallel Manipulator

This paper presents forward and inverse Kinematics analysis of a specific class of series– parallel manipulators, known as 2(6-UPS) manipulators. As Forward kinematics problem of this kind of manipulators is a very difficult problem to solve because of their highly nonlinear relations between joint variables and position and orientation of the end effectors, Numerical methods are the most common approaches to solve. Nevertheless, the possible lack of convergence of these methods is the main drawback. Therefore, artificial neural networks (ANN) with their inherent learning ability as a strong method, was used to approximate the forward kinematics function without any knowledge of manipulator structure. In this paper, two types of ANN models were used. MLP (multi-layer perceptron network) and RBF (radial basis function network) have been used to solve the forward kinematics problem of this hybrid manipulator and results are obtained. Simulation results show the advantages of employing neural networks. Also, according to average percentage error, as the performance index, it was found at RBF gives better result as compared to MLP.