Quadcopter Control System Design and Implementation with LQR assisted by Kalman Filter: An Experimental Study

Abstract: This study stems from the audacious assumption that two bachelor students could successfully construct a functional quadcopter from scratch, covering both software and hardware aspects. The project begins by establishing a foundation for modeling the quadcopter dynamics, including derivation and linearization techniques for implementing a linear quadratic regulator (LQR). To validate the model, a MATLAB simulation environment is developed, providing valuable qualitative insights for hardware selection and a holistic understanding of the quadcopter system. Moving beyond simulation, the implementation phase involves selecting hardware components based on simulation results, consultation, and practical considerations. CAD modeling is utilized for 3D printing, post-processing, and assembly, resulting in a physical drone. Extensive experimentation and programming are conducted to enhance the robustness of sensors, while the challenges posed by radio modules are addressed through message splitting techniques and the development of a software architecture that leverages the computational power of a laptop. With the fundamental functions in place, efforts focus on approximating the idealized model. While the BNO055 effectively approximates six of the twelve states, accurate estimation of the remaining six states, hence position and velocity in all three degrees of freedom, proves challenging. Incorporating a GPS, barometer, and sensor fusion techniques is explored but does not yield satisfactory results due to velocity drift and slow position updates. Although this report does not demonstrate the performance of a fully functional flying quadcopter, it provides practical solutions to the challenges encountered during the construction process. These solutions offer valuable insights and compensate for the initial ambitious goals. The subsequent sections of the report analyze the encountered issues, draw conclusions about quadcopter theory, and propose avenues for future development.