Advanced control techniques for unmanned aerial vehicles

This thesis investigates the development and evaluation of a quadrotor controller based on Koopman theory, and compares its performance with a benchmark PID controller that includes feedback linearization in its inner loop. The main objective is to assess the effectiveness of the Koopman-based controller relative to the traditional PID approach through a series of simulations and physical tests. The evaluation encompasses three main stages: Model-in-the-Loop (MIL) simulations, Hardware-in-the-Loop (HIL) simulations, and physical trials on a custom-built quadrotor drone. Both controllers were tested on two distinct trajectories: hovering and circular paths. Specifically, the Koopman-based controller was tested with varying speeds during the circular trajectory tests to explore its operational limits. The results provide a comparative analysis of the controllers' performance and robustness in different conditions, contributing to the advancement of quadrotor control strategies.