Pitch Optimization via a Learning Algorithm (POLA) for an Aerial Single-Rotor System

Current aerial single-rotor thrust vectoring systems control attitude and angular velocity using a swashplate, which mechanically constrains the blade pitches to sinusoidally vary in phase with and at the frequency of the rotor rotation. While superposed sinusoidal actuation at higher harmonics has shown great benefits, there remains a large search space comprised of sinusoidal, nonsinusoidal, harmonic, and nonharmonic actuation. The complex nonlinear dynamics of rotor aeroelasticity combined with such a vast search space befits the application of machine learning techniques. We equip a rotor with individual blade control (IBC) for unconstrained authority over pitch trajectories. This rotor is integrated with a cyberphysical system where real-world experiment drives the progression of an evolutionary algorithm. By optimizing pitch trajectories to efficiently vector thrust, we gain insight to system dynamics.