Robot Identification using Dynamical Neural Networks
It is nowadays well known that neural networks can model very efficiently complex nonlinear systems. This paper solves the identification problem of a robotic manipulator using dynamical neural networks. More explicitly a dynamic, distributed backpropagation network with two hidden layers and a novice algorithm are used. The network includes dynamic el-ements in its neurons, and this property makes it effective in identifying dynamic nonlinear systems. Simulation results demonstrate the applicability of the approach.
KeywordsNeural Network Hide Layer Synaptic Weight Robotic Manipulator Dynamical Neural Network
Unable to display preview. Download preview PDF.
- IEEE Transactions on Acoustics Speech and Signal Processing special issue on Neural Networks, vol. 36, no. 7 1988.Google Scholar
- D. Rumelhart, D. Hinton, and G. Williams, “Learning internal represantations by error propagation,” in D. Rumelhart and F. McGlelland,edsParallel Distributed Processing,Vol. 1. Cambridge, MA:MIT Press, 1986.Google Scholar
- M. A. Cohen and S. Grossberg, “Masking fields:a massively parallel neural architecture for learning, recognizing, and predicting multiple groupings of patterned data,”Applied Opticsvol. 16, no. l0,pp. 1866–1890.Google Scholar
- J. J. Craig, Introduction to Robotics. Mechanics and Control, second edition, Addison-Wesley Pub. Co., 1989.Google Scholar
- A. Chassiakos, “Distributed control of robot manipulators,” PhD thesis, University of Southern California, LA, CA, 1984.Google Scholar
- P. D. WassermanNeural Computing. Theory and PracticeVan Nostrant Reinhold, New York, 1989.Google Scholar