A Novel State Estimator and Low Level Controller to Enable Nonlinear Model Predictive Control of Tensioned-Tethered Drones

Master thesis defense

Jonas Keplinger

University of Freiburg

Friday, April 24, 2026, 8:30 - 10:00

Building 102 - SR 02-012

Abstract:

In this thesis we develop a controller for flying an unmanned aerial vehicle (UAV) on a tensioned tether in a real-world setup. The controller performance is evaluated using a low-cost testbed built around the Crazyflie quadcopter and the Lighthouse positioning system. First, we develop a linear quadratic regulator (LQR) that runs onboard the drone and stabilizes it around given setpoints. Subsequently, we develop a model predictive controller (MPC) that runs on an external host PC and produces optimal setpoints for the LQR. By combining the two controllers in a hierarchical architecture we are able to track a periodic circular reference trajectory along with a desired tether force. We evaluate the controller performance for a circle with a period of four seconds per revolution and the tether at an angle of 20 deg from the vertical, while the drone exerts a force of 0.1 N on the tether. The proposed controller serves as a foundation for further research on airborne wind energy (AWE) systems with multiple kites. Videos of the experiments can be viewed on YouTube.