Vision-Guided Control Algorithms for Micro Aerial Vehicles

This study presents an image analysis method used in the vision guided control system for Micro Air Vehicles (MAVs). The paper describes a hypothetical model of a MAV located in the GPS-denied unknown environment, somewhere indoors. The model keeps moving autonomously following ‘the track’ marked with corners and other feature points recorded with a monocular camera pointed at the far end of a corridor and slightly tilted down at the angle β (20° - 30°). The flight stability and control are provided with an on-board autopilot that maintains zero pitch and roll angles and constant altitude. The image analysis has been based on the real-time computer vision library - OpenCV (Open Source Computer Vision library - http://opencv.willowgarage.com).