Simulation and Flight-Test Evaluation of Fault-Tolerant Control Allocation Strategies for eVTOL Aircraft

Abstract

Electric vertical takeoff and landing (eVTOL) aircraft typically have more control effectors than controlled axes. This redundancy can improve flight safety by enabling recovery from control effector failures. However, eVTOL aircraft pose unique control allocation challenges because control effectiveness varies across the transition flight envelope, requiring algorithms that can redistribute virtual force and moment commands among redundant effectors under saturation and fault constraints. This thesis presents a geometric evaluation framework for characterizing the force and moment generation capability of an aircraft using the attainable force and moment set (AFMS)-a convex set in the six-dimensional space of generalized force and moment control inputs constrained by saturation and faults. Multiple pseudoinverse-based and optimization-based linear allocation methods are evaluated using the conceptual NASA Lift Plus Cruise eVTOL aircraft in steady-flight and dynamic-flight simulations. In steady-flight simulations, performance is benchmarked using AFMS volume ratio, control effort, and computational burden in nominal and faulted conditions. Dynamic-flight simulations additionally evaluate allocation error, trajectory tracking, and fault-hiding. Finally, a flight-test demonstration on an overactuated subscale fixed-wing aircraft validates real-time executability of the pseudoinverse, redistributed pseudoinverse, and direct allocation algorithms within a PX4-based architecture. Across the steady-flight, dynamic-flight, and flight-test evaluations, direct allocation provided the strongest overall performance, where the redistributed pseudoinverse emerged as the best lower-complexity alternative. Overall, this thesis establishes a scalable, aircraft-agnostic framework for comparing and selecting practical, fault-tolerant control allocation strategies.