Artificial neural networks aided solution to the problem of geometrically bounded singularities and joint limits prevention of a three dimensional planar redundant manipulator

This paper presents a neural network based on a nonlinear dynamical control of a three-dimensional six degrees of freedom planar redundant manipulator. An artificial controller is used for the computation of fast inverse kinematics, and is effective on geometrically bounded singularities and joint limits prevention of redundant manipulators. A comparison between the results of a multilayer back propagation and the radial basis function neural network has been carried out, and the results show that the radial basis function of neural networks is more attractive due to their fast training, simplicity, and convergence rate. The radial basis function neural network has been used to estimate the centrifugal and gravitational effects of the joints, while the end-effector follows a desired path.