A Master of Science thesis in Mechatronics Engineering by Roozbeh Falah Ramezani entitled, "Non-Linear Modeling and Control of Unmanned Air Vehicle," submitted in May 2012. Thesis advisor is Dr. Ali Jhemi. Available are both soft and hard copies of the thesis.
The aim of this thesis is to design and simulate a Dynamic Inversion based autopilot for a fixed wing aircraft. The autopilot provides the aircraft motion stability by commanding the different aircraft control surfaces and achieve a given set of performance specifications. The Dynamic Inversion controller performance is evaluated and tested against classical autopilots. Issues related to stability and robustness are dealt with during the autopilot design. Almost all today's autopilots are design based on linear aircraft models using PID techniques. To insure adequate performance of such simple PID autopilot, flight envelope is divided into many flight modes, each with different autopilot structure. Normally, a linear gain scheduler is used to transition these autopilots from one flight condition to the next. To achieve the performance required, a very complex code is designed and implemented. Such large software overhead makes the testing and certification very costly and very time consuming. In addition, under stringent flight conditions, it is very difficult to guarantee the performance and robustness of the system. Dynamic Inversion controllers are a natural solution to the performance and robustness problem. Dynamic Inversion reduces the code size significantly and simplifies the testing and validation process of the system. So the autopilot can be run on a smaller and lighter hardware. In addition, the stability and robustness of the Dynamic Inversion controller are taken into account during design using mu-synthesis and therefore less testing is required. In this thesis a 6 DOF, 12 states rigid body non-linear model of a small fixed-wing aircraft is developed. Gravity, aerodynamic, and propulsion forces and moments are considered. Linearized models are extracted for a set of given points by trimming the non-linear model about some specified flight conditions. A Dynamic Inversion control-law is derived and robustness and stability issues are addresses during design. Our simulated results show that the performance specification of the system is satisfied and that the tracking of the Dynamic Inversion controller is by far much better than that of a classical PID. Search Terms: UAV, Dynamic Inversion, Autopilot, Non-linear controller, Aircraft