With the advent of low-cost and easily accessible drones, using drone cameras for photography has become more popular than piloting traditional airplanes for ordinary people. When operating a drone from a remote location, its built-in camera is often used to capture images. Unlike traditional airplanes, drone photography is typically done in urban areas, which requires a 3D map that accurately reflects the height of buildings. Multiple perspectives must be used to project the map, similar to 3D computer graphics software. Those views allow for projecting perspectives from arbitrary angle and positioning relative to the drone. Unlike a traditional airplane navigator, the system should display detailed information about small buildings in the city like the navigation maps for automobiles. In the camera view, to avoid collisions, it is necessary to use OpenCV and PyTorch image recognition to roughly understand the 3D shape of buildings in the city and vectorize it in the camera view. Additionally, as in the case of automobiles, the system must warn and automatically control the drone to avoid collisions when it gets too close to the planes or edges that constitute the shape of the buildings. In this paper, we present a prototype system with such a map view and consider the layout and feasibility of the view system.
Map View for Remote Control of Drones
2023-10-01
1178415 byte
Conference paper
Electronic Resource
English