An Integrated Unmanned Aerial Vehicle System for Vision Based Control

Abstract

As the development of computer vision and the popularity of unmanned aerial vehicle, using visual information to control the UAV motion becomes a hotspot. Time-to-contact is one of the concepts that is used to control robot motion such as braking, landing, perching, and obstacle avoidance based on visual information. In this paper, to explore the capability and potential of a direct featureless time-to-contact estimation algorithm for unmanned aerial vehicle motion control, we design an integrated unmanned aerial vehicle system and verify the accuracy of the featureless time-to-contact estimation algorithm. In addition, compressed sensing is combined with the featureless method in time-to-contact estimation to potentially improve the computational speed. The experiment results show that the featureless time-to-contact estimation algorithm, the developed UAV platform and compressed sensing can be readily applied for UAV vision based control.