Vision-Based Autonomous Navigation Of An Unmanned Aerial Vehicle

Abstract

This article is focused on the real-time detection and recognition of the objects by using acamera located on Unmanned Aerial Vehicle (UAV). The detected objects in the image carry informationabout the next points of the flight plan according to which the drone performs an autonomousflight. The drone can recognize the information encoded in the patterns on the ground by the field ofview of the camera, therefore, it can perform flight manoeuvres consequently. The detection andinformation recognition is performed by a single-board computer and it is based on comparing imageswith a predefined Haar cascade. The single-board computer on the drone recognizes a size andspacing of the individual contrast points which determine the partial information about the next flightpath. Afterward, single-board computer transmits the detected information to the Pixhawk controlunit by using the Mavlink protocol in real-time. The aircraft control unit ensures the accurate positioningof the UAV by sequential execution of the commands.