Hamiltonian mechanics and neural network technology for high stabilization of Quadrotor UAV on SE(3)

Abstract

The most important feature of expensive drones is that the flight stability is overwhelmingly high. The big expensive drones can fly safely simply due to their big size and heavyweight. The downside of small drones is that the flight is not stable and is easily affected and this is not only due to the wind but also to the nonlinear approaches to be used for flight control systems and the architecture used. In this paper, we propose a new nonlinear dynamic modeling based on Hamiltonian mechanics the most powerful model in modern mechanics for high attitude stabilization of a quadrotor aircraft, a nonlinear controller is developed on the special Euclidean group SE(3). The fly controller is designed by the neural networks (NNs) control algorithm executed on a microprocessor and makes his calculation in a high-speed calculator unite to achieve the desired tracking for the drone.