Implementation of a Data-Driven Control Method for Unmanned Aerial Vehicles
Yevhenii Kovryzhenko, Nan Li, Ehsan Taheri · 2024
Interest in data-driven control methods is growing due to their ability to leverage vast amounts of data to optimize efficient operation of complex systems. These methods improve decision-making accuracy and adaptability in various applications, ranging from industrial automation to autonomous vehicles. In this paper, we investigate the applicability of data-driven methods for the development of flight control of unmanned aerial vehicles (UAVs). We synthesize robust control laws based on a matrix S-lemma approach. The theoretical foundation of the matrix S-lemma approach is reviewed, and the formulation and solution of the resulting linear matrix inequalities (LMIs) are presented. Its implementation is demonstrated on a six-degrees-of-freedom non-linear UAV system. Our test results in a high-fidelity simulation environment illustrate the ability of the derived control law to stabilize the non-linear UAV system and achieve a closed-loop performance comparable to a model-based linear-quadratic regulator (LQR) design. This research contributes to bridging the gap between theoretical advancements in the field of data-driven control and tangible applications in the field of UAV~systems.