Closed-Loop Data-Driven Attitude Control Design for a Multirotor UAV

Abstract

Data-driven controller design methods allow a fast tuning of controller parameters directly from data, relying on limited prior knowledge of the plant dynamics. In this paper, the problem of tuning the attitude control system of a multirotor UAV is tackled and a data-driven approach is proposed. With respect to previous work, in which open-loop experiments were carried out on a test bench allowing only a single attitude degree of freedom of the vehicle, in this paper data collected in flight, during closed-loop experiments, is used to tune the controller gains. The results, based on experimental work carried out on a small-scale quadrotor, show that a performance level comparable to that of model-based methods can be achieved.