Diseño de un controlador para el seguimiento de trayectorias de un vehículo autónomo no tripulado de Ala Rotatoria.

Abstract

This report presents the design of a controller for trajectory tracking of an unmanned autonomous rotary-wing vehicle. The goal is to optimize the accuracy and stability in trajectory tracking, overcoming the limitations of the pre-existing internal controller in the AirSim simulator. Several scripts were developed in Visual Studio Code, including one for generating random trajectories and another for tracking these trajectories using the internal controller of AirSim. Additionally, a neural network was designed and trained to complement and improve the performance of this controller. The results demonstrated that the integration of the neural network optimized the accuracy and stability of the drone's trajectory tracking, surpassing the limitations of the pre-existing internal controller. Implementing a neural network as a complement to the control system significantly improves the performance of the autonomous vehicle in terms of accuracy and stability in trajectory tracking. Keywords: Trajectory controller, autonomous vehicle, rotary-wing drone, AirSim, neural network, trajectory tracking, artificial intelligence, flight simulation, controller optimization.