System identification modeling of micro air vehicles from visual motion tracking data.

Abstract

This thesis describes a new technique, known as direct trajectory sampling, for extracting aerodynamic coefficients directly from Micro Air Vehicle trajectory data. The experiment is conducted in the Interactive Guidance and Control Laboratory (Aerospace Engineering & Mechanics, University of Minnesota) using the vision tracking system and an aircraft fitted with reflective markers on its wings, fuselage, nose and tail. The technique was developed to study the aerodynamics of small-scale aircraft as the traditional methods using wind tunnels and computational fluid dynamics have produced limited results for this rapidly growing field of research. The indoor environment takes advantage of the stationary atmospheric conditions and as the glider is launched into flight the full trajectory is captured by the tracking system while the aircraft position and attitude vary naturally throughtout the flight. The experimental data is filtered with a Kalman filter to eliminate any data drop-outs and to generate velocities and accelerations. Aerodynamic forces, moments and angles are extracted from the filtered flight data and are used in the calculation of the aerodynamic coefficients. Stability derivatives are determined from the aerodynamic coefficients using a linear regression approach, and these are then refined with a flight simulation model and the technique of maximum likelihood estimation. Once the stability derivatives have been refined, resulting in an identified system, the aircraft simulation model is validated using the weighted mean of each stability derivative. Pleasing results are obtained in several cases but further refinement of the model is required before confident claims regarding the validation can be made.