Evolutionary structured RBF neural network based control of a seven-link redundant manipulator

A method for the identification of complex nonlinear dynamics of a multilink robot manipulator using Runge-Kutta-Gill neural networks (RKGNN) in the absence of input torque information is proposed. The RKGNN constructed using shape adaptive radial basis functions (RBF) are trained using an evolutionary algorithm. Due to the fact that the main function network is divided into subnetworks to represent detailed properties of the dynamics of a manipulator, the neural networks have greater information processing capacity and they can be tested for properties such as positive definiteness of the inertia matrix. Dynamics of an industrial seven-link manipulator are identified using only input-output position and their velocity data. Promising experimental control results are obtained to prove the ability of the proposed method in capturing highly nonlinear dynamics of a multilink manipulator in an effective manner.