Data-Driven Attitude Control Law of a Variable-Pitch Quadrotor: a Comparison Study

In this paper, the problem of tuning a cascade attitude control system of a variable-pitch quadrotor UAV is tackled, comparing two non-iterative data-driven approaches. The first method is the Virtual Reference Feedback Tuning (VRFT) while the second one is the Correlation based Tuning (CbT), both modified in order to tune both the inner and the outer loops by means of a single set of experimental data. These methods allow a fast tuning of controller parameters directly from data, without relying on an accurate knowledge of the plant dynamics. The experimental tests are performed on a variable-pitch quadrotor which operates indoor on a dedicated test bench that allows only the pitch attitude degree of freedom of the vehicle.