Computer vision in remotely piloted aircraft (RPA) to avoid obstacles during flight

Abstract

This paper presents the implementation of an algorithm based on elementary computer vision techniques that allow an UAV (Unmanned Aerial Vehicle), or robot, to identify obstacles in front of it and quickly decide a better way to move in order to avoid them, without demanding complex software or high computational performance, using only a trivial camera and applying no more than two mathematical treatments on images. We implement and test this solution in a simulation environment, although it can be readily applied to UAVs in real life.