Diseño y simulación de un estabilizador de cámara controlado por visión computacional para un vehículo aéreo no tripulado de ala fija.

Abstract

The use of Unmanned Aerial Vehicles (UAVs) currently allows environmental monitoring applications. The use of a camera incorporated in a UAV is essential to carry out this type of tasks, however, due to vibrations of the UAV and the forces of the wind, it is necessary to use a gimbal to stabilize the camera to obtain excellent quality of images. The control of a gimbal is commonly based on an inertial measurement unit (IMU) sensor that detects the movement of the mechanical system and stabilizes it using motors. The gimbal design of this thesis focuses to control stabilization by open-source Computer Vision algorithms that will allow future researchers to adapt it to not only stabilize a camera but also to perform more applications. The algorithm takes outdoor images as input data and detects the sky line that is aligned with a reference image by controlling the roll rotation angle, which is the angle of rotation that presents the greatest instability during a flight test. The gimbal design is based on a 3-link mechanism that allows a Raspberry camera to rotate in roll and pitch using two servo motors that will be controlled by an NVIDIA Jetson Nano board. The structure of the gimbal is designed for a bullet aerodynamic profile with a low drag force does not affect the flight of the UAV. The design also seeks simple manufacturing and low weight to accommodate the maximum payload of the UAV