Robust Visual Servo Control With Visibility Constraints for High-Dynamic Landing of Quadrotors

High-dynamic and high-precision landing of visual servoing quadrotors poses significant challenges, primarily attributed to the low-frequency feedback characteristics of visual sensors and the physical coupling between the field of view (FOV) and quadrotor attitude variations. This paper presents a composite nonlinear feedback control framework based on visual servoing (PBVS-CNF) for the autonomous landing of quadrotors on rapidly moving platforms. First, an adaptive interpolation-compensated extended Kalman filter (AICEKF) is designed to compensate the measurement delays and obtain high-frequency updates of visual positioning information. Subsequently, a composite nonlinear feed back performance-optimized controller based on a pixel repulsion function (PRF) is developed to maintain continuous target visibility during high-dynamic landing tasks, while enhancing the vehicle's response speed and terminal landing accuracy. A compensation function observer (CFO) is designed to estimate full-course landing uncertainties via multi-source state fusion, exhibiting higher accuracy and faster response speed than conventional observers. Finally, simulations and experiments validate the effectiveness of the proposed method.