Application of neural network training in forward kinematics simulation for a novel modular hybrid manipulator with experimental validation

作者：Rahmani Arash^*; Ghanbari Ahmand

来源：Intelligent Service Robotics, 2016, 9(1): 79-91.

DOI：10.1007/s11370-015-0188-8

• 2-(6UPS) manipulators
• Image processing
• Kinematics simulation
• Nonlinear multivariable system
• Wavelet neural network

This contribution addresses forward kinematic solution of modular hybrid manipulator which includes two similar Stewart mechanisms in serial form, known as 2-(6UPS) manipulator. First, using geometrical and vectorial analysis, mathematical model of kinematic analysis for 2-(6UPS) is extracted. Due to highly nonlinear characteristic of kinematic model of 2-(6UPS), which is related to complicated configuration of manipulator, forward kinematic solutions of mechanism is so difficult. Therefore, we proposed artificial neural network based on wavelet analysis to resolve forward kinematics of 2-(6UPS). Also, we used wavelet neural network (WNN) to approximate specific trajectory in circle oscillates in z direction and spiral with semi-cardioid base paths for mid and upper platforms movement respectively. Comparison between the results of proposed network and closed form solution of kinematics for 2-(6UPS) shows proper performance of proposed network in <1% error. Also, WNN results are verified by experimental results which are obtained by image processing.

• 出版日期2016-1