Icing Detection, State Estimation, and Control of Fixed-Wing Drones

Abstract

Reliable operation of unmanned aerial vehicles (UAVs) in uncertain and adverse conditions

remains a critical challenge, particularly in the presence of icing, disturbances, and dynamic in- teractions in multi-agent systems. This thesis develops an integrated framework that combines

probabilistic estimation and nonlinear control strategies to enhance performance and robust- ness. An approach based on the particle filter (PF) is employed to improve state estimation ac- curacy and detect icing-related faults by analyzing variations in system parameters. For robust

trajectory tracking in uncertain flight conditions, the proposed control scheme combines a high-order sliding mode observer (HOSMO) with the super-twisting algorithms (STA), effectively managing disturbances and model variations. At the multi-agent level, a distributed control strategy is introduced, utilizing finite-time observers and controllers within a leader–follower

structure to enable fast and coordinated group behavior. The thesis demonstrates the effec- tiveness of the proposed methods in enhancing fault detection capabilities, control robustness,

and ensuring reliable multi-UAV coordination.