Unified Guidance & Control Framework for eVTOL Aircraft

Abstract

Autonomous electric vertical takeoff and landing (eVTOL) aircraft, particularly transitioning architectures like tiltwings, demand tightly integrated autonomy stacks capable of bridging thrust-borne hover and aerodynamic cruise. This dissertation presents a unified guidance, feed-back control, and control-allocation framework designed to manage the severe nonlinearities and actuator redundancies inherent to morphing geometries across the full flight envelope.

First, the research derives exact, singularity-free analytic differential flatness mappings for multirotor, fixed-wing, and tiltwing configurations. Despite differing force-generation mechanisms, all platforms share a common four-dimensional flat output set comprising inertial position and heading. By employing sequential coordinate transformations and a novel wing-frame force projection, the mappings decouple lateral and longitudinal dynamics, yielding analytically exact feedforward force and moment commands from trajectory derivatives without online numerical integration that are dynamically (but not necessarily operationally) feasible.

Second, a unified cascaded feedback control architecture is developed to track these dynamically feasible references. For the tiltwing configuration and to address the shifting physical paradigms of flight, translational errors are projected into a wing-aligned control frame, mathematically eliminating multi-variable gain cross-coupling at intermediate tilt angles. A continuous, regime-weighted blending strategy seamlessly interpolates between underactuated thrust-vectoring in hover and aerodynamic load-factor pitch laws in cruise, ensuring mathematical continuity and preventing integrator windup without requiring auxiliary tuning.

Finally, to resolve the time-varying, nonlinear control effectiveness matrices of over-actuated eVTOLs, a progressive control-allocation framework is established. Moving beyond traditional pseudoinverse methods, the architecture integrates successive linearization with trust-region Gauss-Newton schemes and exact box-constrained quadratic programming (QP). This optimization-based allocator strictly enforces physical actuator saturation limits, prevents directional clipping, and enables dynamic task prioritization during severe morphing transitions.

Validated through high-fidelity simulation of a tiltwing forward-transition sweep, the integrated pipeline demonstrates seamless compatibility across planning, allocation, and tracking stages. By unifying differential flatness, wing-frame feedback decoupling, and robust optimization-based allocation, this work provides a computationally tractable, disturbance-resilient autonomy framework that accelerates the safe deployment of complex eVTOL platforms for Urban Air Mobility.