ABSTRACT:
A ducted fan has been constructed at California Institute of Technology, in order to provide an
experimental test-bed for research and development of different control techniques for
Uninhabited Combat Aerial Vehicles (UCAVs). It has many of the characteristics of existing
flight control systems and its identification experiments have shown that its model is correct
near hover and successful control designs also support this conclusion. This paper presents a
comparison between the application of one of the nonlinear Model Based Predictive Control
techniques (MBPC), which is called Non-Linear Quadratic Gaussian Predictive Control
(NLQGPC), to two different representations of the ducted fan hover model. The paper also
provides an insight into the strengths and weakness of each algorithm and into the types of
problems, which limit the applicability of those algorithms.