Modeling robot dynamics using dynamic neural networks

摘要

This paper is concerned with modeling the complex nonlinear dynamics of a robotic ssytem using dynamic neural networks. The effect of the shape of the slope of the activation function (the node nonlinearity) on the performance of the learning algorithm in training a network has also been investigated. A modified form of the back propagation learning algorithm, with adaptive weights and adaptive slope of the nonlinearity of the neuron has been used. To demonstrate the validity of the proposed method, we apply it to identify the inverse dynamics of a two link manipulator.