An intelligent controller design for quadrotor type uav

In this thesis study, it is focused on the control of quadrotor, which is a common unmanned aerial vehicle, using the Artificial Neural Network, which is an intelligent structure as an alternative to the classical PD controllers. In the study, simulations are carried out with respect to the control of the unmanned aerial vehicle with conventional PD controllers, and the behavior of the unmanned aerial vehicle is observed for the scenario involving noise. In the following, PD controller simulations are made for 256 different trajectories, then PD controllers are replaced by feedforward neural networks which are trained by using these training data regarding roll, pitch, yaw, and z in the inner loop. Then, simulations are implemented using artificial neural networks trained by a single training trajectory. From the results obtained, it is shown that artificial neural network controllers can be substituted for classical PD controllers for quadrotors.