Drone Position Tracking System based on Object Detection and ArUco Marker for Autonomous Navigation Applications

In an autonomous navigation system, the drone needs to know its position continuously to navigate properly to the destination. In indoor environments, localization using GPS (Global Positioning System) often results in poor accuracy due to various factors such as building structures that affect GPS signal reception. In this paper, we propose an indoor drone position-tracking system based on visual sensing from the drone camera. Two types of artificial visual markers such as ArUco markers and 3D object markers are used as references to determine the relative position of the drone. The estimated drone position is obtained by computing the translation vectors of the artificial visual markers in the vision of the drone camera. The estimation results with the combination of object detection and ArUco markers show an average accuracy of 10 cm with a range of 0.1 - 8 meters.