12th IEEE International Workshop on Metrology for AeroSpace, MetroAeroSpace 2025, Naples, Italy, 18 - 20 June 2025, pp.274-279, (Full Text)
Quadrotors are becoming more prevalent for inspection, delivery, and industrial tasks, but their underactuated design makes them susceptible to actuator faults. Faults can range from reduced rotor effectiveness to complete rotor loss, resulting in catastrophic performance degradation or crashes. Existing approaches are generally linear approximations or partially fault tolerant, limiting their applicability under extreme conditions or external disturbances. In this paper, a Nonlinear Model Predictive Control framework is combined with an Incremental Nonlinear Dynamic Inversion approach to control partial and full rotor failures. With the explicit handling of the rotor thrust constraints and the addition of the wind dynamics, the controller proposed attempts to provide stable flight as well as agile maneuverability and eventually ensure a controlled descent and safe landing even in fault conditions. Our methodology goes beyond one rotor failures to mitigate simultaneous partial or complete loss of two opposite rotors while maintaining limited control authority and facilitating stable maneuver at high angular rates. Moreover, with the incorporation of aerodynamic mismatches and wind disturbances within the controller, we prove robust performance both at low speeds and high speeds of flight. Accordingly, the method provides a full solution for real-time fault tolerance, extending quadrotors' flight envelope in unstructured spaces.