Robust trajectory tracking control of an underactuated control moment gyroscope via neural network–based feedback linearization

In this document, an underactuated two degrees–of–freedom control moment gyroscope (CMG) is studied. Specifically, the problem of trajectory tracking in the non-actuated joint is addressed. The feedback linearization technique is used to design a model-based controller. Then, an adaptive neural network–based scheme is designed to add robustness with respect to model uncertainties. Studies on the internal and output dynamics are presented. The introduced theory is validated by means of real–time experiments. The comparisons among a linear controller, a cascaded PID–PID scheme, and a known adaptive neural network controller are presented to assess the performance of the novel robust controller given in this work. Better tracking accuracy is obtained with the introduced approach.