3D Environment Map Reconstruction With Aerial Camera On A Drone

Abstract

If an accident happens on the drone flight, the flight recorder system is helpful for finding causality of crush. However, the general flight recorder system can not restore flight route in detail because it is not monitored environments by the camera. In this paper, we will propose to restore a system which uses 3D scenes using a monocular camera. Constructing a scene around where a drone flew the flight route from camera images/video using SLAM (Simultaneous Localization and Mapping). After constructing the route, we will convert point clouds to a voxel scene for noise or drawing time reduction and making easier to understand than point cloud. Experimental results show that our proposed system is effective to reduce noise and drawing time, however, difficult to compare with easy understanding because LSD-SLAM cannot recover entire 3D environment.