Design of a Multiple-IMU Navigation System and preliminary study of Machine Learning applications

One of the key features of an aircraft is the Navigation System, which is responsible for the evaluation of the actual localization during a mission. Moreover, the Navigation System is the interface of determination of position and velocity, and all the aircraft parameters thanks to the on-board sensors. Navigation is usually carried out thanks to the integration of several sensors and technologies. The Global Navigation Satellite System (GNSS) and the Inertial Navigation System (INS) are widely used, combining the precision in estimating the position and velocity of the former with the high update rate of the latter. The integration of GNSS and INS yields high performance and accuracy in a nominal scenario, under ideal conditions. However, the GNSS is susceptible to a range of vulnerabilities, particularly in conflict zones where it can be jammed or spoofed. Under these conditions, we speak of GPS Denied Navigation. As a consequence, the Navigation System is required to rely on a self-contained system which is able to perform with minimal outside information. The purpose of this thesis is to focus on Inertial Navigation and the evaluation of its performance through the implementation of a Multiple-IMU Navigation System. The integration of two Inertial Measurement Units (IMUs) with various technologies and features is discussed, studying the system’s drift, and thus the overall aircraft’s navigation performance, with a sensor management implementation. Various scenarios are simulated, considering several trajectories and different GPS availability ranges. Simulation results are reported and discussed, comparing the effects of different fusion algorithms customized for this application, and evaluating the performance comparison of a Single-IMU versus a Multiple-IMU System. In conclusion, a framework of future trends for Machine Learning applications in a Multiple-IMU Navigation System is presented, through a preliminary study and comparison with the more traditional methods of systems integration.