Non-Linear Control of Unmanned Quadcopters
Dimitris Nikitas
Abstract
In this work, a non-linear mathematical model is derived through kinematic and dynamic analysis and elemental blade theory. Firstly, linear control strategies based on the linearized model, such as pole placement and LQR, are tested. Furthermore, two non-linear cascaded controllers are obtained through Back-stepping design. These nonlinear strategies’ performance is compared to that of the linear ones. Then, the least time problem for Rest to Rest transition is defined using Pontryagin’s Maximum Prin-ciple for quadrotor dynamics. A novel near optimal controller is designed which solves the problem for either planar or altitude motion, requiring only a minimum of com-putational effort, in contrast with most time optimal controllers. The controller is closed loop and the idea of cascaded Bang-Bang control is proposed. A simulator and a graphic environment were developed in MATLAB for algorithm testing. An intent of this thesis is to provide a reference point for a new potential thesis in the lab related to the execution of the experimental part.