Robust Tuning of Geometric Attitude Controllers for Multirotor Unmanned Aerial Vehicles

In recent years there has been a significant body of literature proposing nonlinear attitude control laws for small-scale multirotor unmanned aerial vehicles (UAVs), motivated by the high maneuverability of these platforms. While tracking trajectories characterized by fast and large attitude changes makes the control problem intrinsically nonlinear, most of the works proposing nonlinear designs is concerned with establishing their stabilizing properties, often deduced by referring to simplified dynamic models, but limited attention has been devoted to performance. As a consequence, less satisfactory results than expected are typically achieved in experiments and the controller gains must be adjusted with trial-and-error procedures to obtain good performance. This paper proposes a model-based tuning method that exploits the cascade structure of the attitude dynamics and that needs only single-axis identified linear models of the angular velocity dynamics to be applied. The tuning of the controller gains is carried out on the linearized closed-loop system with structured H∞ synthesis that allows one to enforce robustness against model uncertainty in a systematic way and to achieve a desired level of performance in nominal conditions. The approach is validated by tuning the gains of a novel Proportional/Proportional Integral Derivative (P/PID)-like cascade, which has been developed in the framework of geometric control theory. A thorough analytical comparison of the proposed design with a geometric Proportional Integral (PI)-like controller borrowed from the literature is complemented with experiments conducted on a small quadrotor UAV.