Randomized Approaches for Control of QuadRotor UAVs

A systematic mixed deterministic-randomized approach for filter design is presented. This method has two advantages: (i) it is a computational efficient algorithm for the deterministic parameters and (ii) we can design fixed order controllers because we follow a randomization approach. Different filter forms are selected to prove the effectiveness of this algorithm and the obtained results are applied to a multi rotor UAV. Both linear and experimental models are analyzed. For the linear case a comparison with a LQR controller is proposed to prove that with this filter the designed controller permits to minimize the noise due to the coupling effects between the different control axes. This design is useful to reduce the set up time of an adaptive controller filter if different maneuvers or aircraft parameters are considered.