Modelling and control of vertical landing rockets

This thesis focuses on a research initiative initiated by SpaceX and Blue Origin, aiming to develop cost-effective rocket technology through vertical landing experiments. The study aims to design a controller to simulate the flight and vertical landing of a rocket, thereby creating a dynamic model for the rocket. In the process of deriving the dynamic model of the rocket's motion equations, various control systems such as gimballed engines, cold gas thrusters, grid fins, and traditional fins will be mathematically modeled using Simulink, a MATLAB-based graphical modeling environment. At the core of the study are comprehensive comparisons of PID (Proportional-Integral-Derivative) controllers, LQR (Linear Quadratic Regulator) controllers, MPC (Model Predictive Control) controllers, and RL (Reinforcement Learning) algorithms in the context of rocket control. The study delves into the simplicity of the PID controller, the optimization principles in LQR design, and the complexity of the MPC controller, focusing on their effectiveness and limitations in successfully controlling the rocket. This thesis stands out as a critical step toward the development of cost-effective rocket technology. The obtained results provide a crucial foundation for future designs of rocket control systems, holding the potential to shape advancements in this field.