Model predictive controller implementation to a multicopter for stability augmentations

Nowadays, multi-rotor systems are widely used in military, civil and academic applications thanks to their cheap, simple and suitable structures for use in different areas such as surveillance and load carrying. On the other hand, thanks to processor and sensor technologies' progress, modern and high computation power-requiring control methods, such as model predictive controller, have become applicable even on these small systems. This thesis study aims to improve the stability and performance of a multi-rotor system with a model predictive controller. The thesis study can be summarized in six main processes. During the study, firstly, a system model consistent with system identification was obtained. Subsequently, this model was verified with flight tests. In the second part, a simulation model was established using MATLAB-Simulink software and, a model predictive controller was designed with "MPC Designer". This controller has been tested in the simulation environment and compared with the reference PID controller. After satisfactory performance results were obtained, C code was generated with "Embedded Coder", it was added to the existing autopilot software, and software in the loop tests were completed. After this step, the generated code was implemented on the Jetson NANO board and the hardware in the loop tests were started. After accomplishing sufficient performance in these tests, flight tests were started. Finally, the system was tested in a real application and compared with the reference controller. Keywords: Model Predictive Control, System Identification, Multirotor, Quadcopter, Software in the Loop, Hardware in the Loop