Celestial Rover Localization

This thesis presents the development and validation of a star tracker based celestial navigation system, called a Digital Star Sextant (DSS), to provide a complete geolocalization solution that does not depend on external aids, such as orbiting satellites, or dead reckoning. Current DSS research focuses on calibration but has not justified cost function selection, explored more suitable measurement models, or assessed calibration parameter observability. Current literature also has not addressed the fundamental shortcomings in inclinometer technology beyond identifying it as a potential inhibitor for achieving useful DSS performance. This thesis establishes data acquisition requirements for calibrating a DSS based on the choice of calibration parameters. A trend in literature has been to conduct calibrations without and instrument to provide heading truth, meaning heading is both measurable and unobservable in calibration. In line with this trend, this thesis introduces a novel DSS measurement model which is deterministic and avoids the otherwise necessary implicit use of heading. The shortcomings of wide range of motion inclinometer technology are addressed, achieving and validating a DSS-enabling accuracy for wide range of motion biaxial inclinometers, marking a state-of-the-art improvement in accuracy. Specifically, an inclinometer calibration for a ±50◦ range of motion inclinometer was validated to within 0.0038◦ based on a sampling resolution predicted to achieve 0.004◦. Singular Value Decomposition of the DSS calibration parameter Jacobian was used to confirm observability of parameters for multiple cost function formulations. Experimental data acquisition for calibration was intentionally limited to confirm predicted parameter redundancies. Finally, an analytical covariance model for the proposed measurement model was validated using a Monte Carlo simulation as well as experimental data. The model predicted errors of 361 m while the experimental errors were 349 m. Together, these results expanded the theoretical framework for designing and conducting DSS calibrations. By providing a mathematical framework which is in-line with existing trends, the work presented in this thesis provides a common foundation for other researcher to build upon. In this way, algorithms and improvement made to DSS system can easily be shared among researchers and propel the development of mission valuable DSS systems.