Robust control for aerodynamic steering of a reusable launch vehicle: an LQG/LT approach with wind disturbance rejection

Abstract

Facing rising costs in space exploration, the aerospace industry has embraced reusability as a transformative solution. A paradigm shift pioneered by SpaceX in 2015 and now adopted globally. This innovation introduces multidisciplinary complexities, particularly in controlling reusable boosters and first stages during atmospheric re-entry and descent. Precise attitude control and aerodynamic manoeuvring are critical to guide these components along predetermined trajectories while ensuring stable deceleration. By leveraging advanced control systems and aerodynamic design, modern launch vehicles achieve recovery and reuse, dramatically reducing mission costs. This dissertation focuses on attitude dynamics and explores a Linear Quadratic Gaussian (LQG) solution for the controlled aerodynamic descent of a reusable launcher using the Loop Transfer Recovery (LTR) method. This technique is based on the fact that the Linear Quadratic Regulator (LQR) using state-variable feedback has specific guaranteed robustness properties. The LQG, like the association of LQR and the Kalman Filter (KF), is tested on a decoupled mathematical model of a launcher vehicle based on the prerequisites of the reusable rocket demonstrator CALLISTO, Cooperative Action Leading to Launcher Innovation in Stage Toss-back Operations, and the existing literature in the field. Using LQG/LTR as an optimal and stochastic control association to propose a robust control of position by direct action on the whole attitude of the vehicle body described by Euler angles, this work aims to investigate the stability and robustness performances of the latter for the control design of the aerodynamic descent phase of a reusable rocket. Therefore, the control structure designed on different cases and the possibility of in-flight events of the rocket in its controlled aerodynamic descent are discussed. Also, different correction strategies are used to refine the initially set performance objectives, like the predictive and weighted guidance trajectory.