Automatic guidance of an aircraft using predictive control in a visual servoing scheme

In this paper we have designed and implemented a predictive automatic flight controller for an airplane that takes as input the estimated pose obtained from the visual data and provides control outputs for an inner-loop augmented aircraft. In order to make the predictive controller suitable for flight control, it had to be modified. The most important modification was the introduction of the short term reference trajectories, linking the time-based and the position-based control strategy. Furthermore a controller switching mechanism was introduced to enable the aircraft to fly at different flight conditions. The parameters of the controller have been derived by experimental optimization. The tests have shown good overall tracking and disturbance rejection properties; the aircraft is able to perform various maneuvers and the touchdown in all the tested weather (wind, turbulence) conditions.