Application of neuro-Kalman filtering to attitude estimation of platforms and space vehicles

In this paper the on going work concerning the application of a particular hybrid neural network and Kalman filtering approach to the attitude estimation problem is presented. A neural network architecture for implementing observers and estimators has been developed. A synergistic approach is adopted not only in learning the weights but also in the development of the macro structure of the neural network. Based on this architecture extended Kalman filtering and neural network learning are combined and implemented to the attitude estimation problem