Automated Coastal Surveillance with Drones Using Gaussian Process Regression Analysis

Authors

  • Hyun-Kyu Jee1 Official, Office of Planning and Budget, Taean-gun Office, Guncheong-ro, Taean-Eup, Taean-Gun, Chungcheongnam-do, 32144, Korea, Korea, Republic of
  • Jae-Yong Lee Professor, Department of Unmanned Aircraft Systems, Hanseo University (Taean Campus), Gomseom-ro, Nam-Myeon, Taean-Gun, Chungcheongnam-do, 32158, Korea, Korea, Republic of

Keywords:

drone, waypoint, tide, GNSS, GPR(Gaussian Process Regression).

Abstract

Background/Objectives: The west coast of South Korea requires an automatic search method that reflects tidal information, because many marine accidents have occurred due to the complex coastline and fast rising tide.
Methods/Statistical analysis: To prevent accidents, drones should automatically search coastlines that change depending on the tidal flow. Thus, coastline data of the west coast were collected with the Global Navigation Satellite System (GNSS), and Gaussian process regression analysis was applied to reflect its latitude and longitude by tide level in real time. Namely, a waypoint where a drone can always search the coastline was predicted by learning with a non-parametric-kernel-based probabilistic model.
Findings: The LITCHI platform was utilized in setting the waypoint, so that the public can use it easily. Mission Hub linked to the web-enabled GNSS waypoint flight by using DJI’s Mavic2, which does not support its own GNSS waypoint flight. Users who can access using Mission Hub can perform, and have real-time multiple control of missions, such as automatic waypoint setting, flight control, and piloting. Additionally, searching for waypoints at night, or in invisible areas where piloting is difficult, is possible, increasing the efficiency of the mission. By synchronizing the tide forecast table of the Korea Hydrographic and Oceanographic Agency with the drone’s take-off time, the changing coastline was confirmed to be searched without error. That is, the drone repeatedly searched the coastline through the standard waypoint predicted by Gaussian process regression analysis. It was confirmed that while the drone is flying, the coastlines that have different shapes depending on the tide level can be observed in real time within the field of view (FOV) range of the camera.
Improvements/Applications: Coastline information collected through automatic flight can be used for marine accident prevention, rescue, and a cooperative system with other organizations.

Downloads

Published

2023-06-04

How to Cite

Hyun-Kyu Jee1, & Jae-Yong Lee. (2023). Automated Coastal Surveillance with Drones Using Gaussian Process Regression Analysis. Utilitas Mathematica, 120, 287–300. Retrieved from http://utilitasmathematica.com/index.php/Index/article/view/1643

Citation Check