Frequency domain model identification and loop-shaping controller design for quadrotor tail-sitter VTOL UAVs

In this paper, we present a practical and systematic method of model identification and controller design for a quadrotor tail-sitter UAV that is susceptible to significant model uncertainty, various flexible modes and complicated aerodynamic damping effect. Flight experiments are designed and carried out to produce a frequency domain model of the aircraft, from which several flexible modes are observed. This system identification method is well suited to the modeling of complex UAV systems, and allows for the rapid update of aircraft models when mechanical changes are made to the aircraft configuration. Besides this, it clearly captures the flexible modes of an aircraft, and enables explicit analysis of them. Guided by this, the controller design, optimization and limitations can be easily seen. Based on the Bode diagram of the identified aircraft model, loop-shaping techniques are applied to design the feedback controllers for all channels (pitch, roll and yaw) of the angular rate loop, and the system stability and robustness are analyzed. We believe that these frequency-domain-based modeling and controller design techniques are promising to qualitatively characterize the aircraft's robustness and performance, as well as to achieve reliable and safe operations of UAVs in actual environments where both model uncertainties and external disturbances often take place. Finally, simulation and experiments are presented to validate the effectiveness of this framework and to show the controller's performance.