Attitude and position control of a quadrotor using onboard vision system

Thesis Type: Postgraduate

Institution Of The Thesis: Middle East Technical University, Graduate School of Natural and Applied Sciences, Turkey

Approval Date: 2019

Thesis Language: English


Consultant: Ulaş Yaman


In this thesis, localization and trajectory following of a quadrotor is obtained with attitude and position control along with the help of an onboard monocular camera. Two control strategies are adopted for this aim, which are well-known PID and nonlinear backstepping controllers. For the PID algorithm, a cascaded structure is preferred so that angular rates are regulated as the inner loop of the attitude controller. At the outer loop of this system, angles are stabilized by producing rate references to the inner loop. PID is a proven method to control nonlinear systems, giving satisfactory system performances. Linearised or simplified system models are generally used to design PID controllers, and implementation is usually straight forward. On the other side, backstepping controller is selected since it does not cancel nonlinearities of the system model. The aim of this method is to obtain the final inputs by creating virtual system inputs in a recursive manner. Since it directly uses system equations of motion, nonlinearities like coupling between axes and actuator frictions are comprised by the controller. This also enables better performance in disturbance rejection. If the strategy is followed properly, backstepping gives satisfactory results. To be able to design these controllers, quadrotor system is modelled within this study. Equations of motion are obtained and quadrotor behaviour is simulated by using Simulink. The responses of the controllers to given references are compared with the usage of this model representation. A real platform is formed for the verification of the proposed control methods. The quadrotor uses a Pixhawk board which includes the main control algorithms. These algorithms are formed in Simulink and Embedded Coder is used to trasfer the program to the Pixhawk. For the localization operation, RaspberryPi board is selected and also used with Simulink application. From the images obtained by the Raspberry, position errors are propagated so that attitude controllers on the Pixhawk can regulate the errors down to zero for stabilization. Serial communication is preferred as the connection interface and hardware and software schemes are explained along this study. As the main aim orients, designed localization and path tracking capabilities are shown on the target platform.