Obstacle avoidance and distance measurement for unmanned aerial vehicles using monocular vision

Abstract

Unmanned Aerial Vehicles or commonly known as drones are better suited for "dull, dirty, or dangerous" missions than manned aircraft. The drone can be either remotely controlled or it can travel as per predefined path using complex automation algorithm. To make it completely autonomous, the most challenging problem faced by UAVs is obstacle avoidance. In this paper, frontal obstacles are detected using monocular vision by extracting features using Computer Vision algorithms like Scale Invariant Feature Transform (SIFT) and Speeded Up Robust Feature (SURF). Distance of the obstacle from camera is calculated by measuring the pixel variation in consecutive video frames. To meet the defined objectives, designed system is tested with self-developed videos which are captured by DJI Phantom 4 pro.