Design and Implementation of a Rocket Launcher Hybrid Navigation

Abstract: Rocket Factory Augsburg (RFA) a German New Space Startup is developing a three-stage rocket launcher aiming at LEO/SSO orbits. A fundamental responsibility of the GNC team is the development of the rocket navigation algorithm to estimate the attitude, position, and velocity allowing the guidance and control loops to autonomously steer the rocket. This thesis focuses on the analysis and design of a Hybrid Navigation system able to satisfy the various necessities of a launch vehicle, such as delay compensation and GNSS outages. The navigation architecture was chosen to be a Closed Loop, Loosely Coupled, Delayed Error State Kalman Filter thanks to the proven capability of COTS receivers to autonomously provide a consistent PVT solution throughout the flight. A preliminary analysis used a reference trajectory to evaluate the effect of the sensor grade on inertial performances and choose an appropriate integration scheme. The filter’s system model was explored using approximate analytical results on observability. The developed navigation module was then tested within a Monte Carlo simulation environment by perturbating the sensor parameter in accordance with the sensor datasheet. As a further verification, the modeled IMU output was compared to the engineering model, to assure that the simulation result would yield conservative errors. Due to concern over the visibility of GNSS satellites during flight, a simplified Almanac-based GPS model has been developed, proving that enough satellite visibility is available along the trajectory. The estimation error was compared with the filter’s estimated covariance and found well within the bounds. Through the study of the covariance evolution, it was determined that given the reference dynamics, the sensor misalignments are the least observable states. Realistic signal outages were introduced in the most critical flight intervals. The filter was indeed found to be robust and the tuning proved to be adequate to capture the dead reckoning drift. Finally, the entire navigation module was deployed onto the avionics engineering model, including the flight computer, IMU, GNSS, and antennas, in a configuration equivalent to flight. The navigation module was then tested to ensure that the execution was in performance under severe multipath errors and prolonged GNSS outages with the covariance estimates correctly covering the uncertainty.