Implementation and comparison of linearization-based and backstepping controllers for quadcopters
Tutor / director / evaluatorTomé Antunues, Guerreiro
Document typeMaster thesis
Rights accessOpen Access
In this work, two approaches to the control of a quadcopter are followed. The rst approach resorts to linear quadratic control (LQR) techniques and is based on the linearization of the quadcopter dynamics. Motivated by the fact that this linearization results in decoupled dynamics for the longitudinal, lateral, height and yaw axis, the LQR controllers can be designed separately. Moreover, the controllers for the longitudinal and lateral dynamics exploit the cascaded structure of the model. The second approach resorts to non-linear control and exploits the fact that the full non-linear model of the quadcopter also has a cascaded structure: the torque inputs control the angles which in turn determine the forces which drive the position states. The approach is based on a widely used non-linear control design technique for cascaded systems known as back-stepping. Simulations of the two approaches are carried out and conclusions are drawn on the pros and cons of each approach.