|
Abstract:
|
Abstract : This master thesis deals with the modeling and control of a small scale birotor tiltrotor unmanned aerial vehicle (UAV). A tiltrotor is characterized by a mechanism that tilts the aircraft's rotors in order to control the flight. An UAV with such characteristics is being developed by this author and other researchers in the scope of the project named ProVANT. The developed UAV prototype is used in this work to obtain experimental results. This kind of system can be characterized by its underactuated, highly nonlinear, and coupled dynamics. Instead of using a dynamic model available in literature, this work proposes a more accurate model considering the UAV as a multibody system. By doing so the tilting angles become generalized coordinates and the tilt mechanism dynamics are naturally added to the model, as well as the coupling between the bodies. The result is an eight degrees of freedom model obtained through Euler-Lagrange formulation. The path tracking problem is solved here with linear optimal controllers for the full model, instead of the classical approach of cascade control for the translation and rotation subsystems. The developed controllers are linear quadratic regulators, a H1 controller and a multiobjective H2=H1 controller, all with LMI formulation. A nonlinear backstepping controller taken from the literature is implemented in order to be compared with the designed controllers. In addition, controllers for the hovering problem are also designed to be used in experiments with ProVANT's tiltrotor. They reduce the complexity of the initial experimental flights, focusing not only in the validation of the control system, but the complete project, including its electronics, mechanical design, and additional software. Such experiments are presented and discussed in details along this work. The work also addresses how flight-related information are gathered and processed. This includes the design of a nonlinear complementary filter for the attitude estimation that works with data acquired from the UAV sensors.<br> |
