This thesis details the development and validation of a robust automatic landing system for a 25 kg fixed-wing unmanned aircraft. It enables automatic landings on unpaved ground and without external infrastructure. The system is capable of landing on non-straight-line approach paths such as curved spline paths and helix-shaped curves to facilitate landings on sites that may otherwise be unreachable due to obstacles or legal restrictions. The thesis contributes a spline path-following algorithm that convinces by simplicity, flexibility, and precision. The entire development is carried out according to the model-based design paradigm. A high-fidelity nonlinear model of the aircraft is developed using comprehensive system identification. It is used to design all controllers using mixed sensitivity H_{infty}-control and classical loopshaping. The automatic landing system's performance and robustness are validated in 150000 Monte-Carlo simulations and real-world experiments. Finally, 36 real-world landings on unpaved ground are performed in varying environmental conditions including noteworthy crosswind, upwind, and turbulence. Detailed results of exemplary flight test experiments are presented. In addition to classical straight-line approaches, landings on curved-path, such as a helix, are successfully demonstrated. The experiments prove the flexibility, robustness, and reliability of the developed automatic landing system. It reduces the probability of human errors and does not require external infrastructure. Its transparent design together with the robust and flexible control strategy simplifies adaptation to other aircraft and provides good supervision for operators.weiterlesen