Micro air vehicle control design : a comparison of classical and dynamic inversion techniques

Abstract

Micro Air Vehicles (MAVs) are envisioned as autonomous aerial robots capable of flying into an urban environment to perform reconnaissance, targeting, and remote sensing. These vehicles are significantly smaller than current aircraft, with maximum dimensions of 15 centimeters, presenting challenges to vehicle packaging and vehicle control. This thesis describes the design and analysis of control algorithms for a rotorcraft MAV using classical and dynamic inversion approaches. Classical control algorithms were developed based on linear, single input/single output analysis and were successfully implemented in hover, climb, and forward flight simulations. A dynamic inversion routine was then applied to the inner-loop control, providing improved inner-loop performance compared to the classical approach for nominal flight conditions. Finally, the two approaches were compared on the basis of their performance, disturbance rejection, and robustness. Dynamic inversion generally outperformed the classical approach for non-linear, state-coupled flight conditions, but suffered when subject to unmodeled dynamics and modeling errors.