[Defense Column] How to Detect North Korean Submarines

[Agency for Defense Development] In the military logistics sector, aerial platforms such as unmanned aerial vehicles are utilized in various fields including real-time reconnaissance and surveillance, electronic warfare, and combat damage assessment. Among these, the ultimate goal of remotely autonomous attack-type aerial platforms is the precise strike of targets. However, even though image analysis technology for detecting and tracking targets is advancing daily, accurate target identification based on the system’s decision logic in battlefield environments requires a much higher level of technical skill than in general environments. In fact, remote control in emergency situations such as changing targets on the battlefield can be directly related to system malfunctions.

To respond to such diverse variables and accurately identify targets, operator intervention inevitably occurs. Man-In-The-Loop (MITL) is a method that allows operators to intervene in aerial platforms and conduct real-time mission control during operations. In a Man-In-The-Loop system, the operator confirms the target image appearing in the video received from optical/infrared (EO/IR) cameras and designates the aiming point of the attack-type aerial platform through the operator’s monitor. The attack-type aerial platform then performs the mission of neutralizing the designated target according to the operator’s command.

Man in the Loop System (Source: Agency for Defense Development)

▲ More Accurate Through Network-Based Systems... The Evolution of Man-In-The-Loop = Recently, as the possibility of applying AI technology to future weapon systems has significantly increased, Man-In-The-Loop systems are evolving. The traditional method, where operators directly designated strike points, is shifting to a network-based environment that utilizes ground computer processing capabilities to suggest strike point locations for the flying vehicle.

Previously, in situations where it was difficult to visually identify target objectives, there was a high dependency on expert judgment. However, by utilizing a network-based Man-In-The-Loop approach, the accuracy of strike points can be improved through real-time information collection and enhanced ground detection and tracking technologies. Furthermore, it becomes possible to execute strike point commands for multiple attack-type aerial platforms with minimal operator intervention.

▲ Applicable Even to Difficult-to-Identify Maritime Targets = Maritime objects are among the targets that aerial platforms find difficult to identify. When detecting and tracking maneuvering maritime objects through the platform’s IR (infrared) camera, noise in the video rapidly increases due to light reflection on the sea surface as the attack platform approaches the water. Additionally, when multiple maneuvering objects form a cluster with similar size, shape, and directionality, sensors with lower shape information levels, such as infrared imaging, cannot guarantee the detection and tracking performance of the intended target.

Therefore, to reduce the impact of noise and classify target objects when clustered objects are input into the video, the Agency for Defense Development developed a technology that separates the background and limits the search range through a background removal algorithm applied to the input video. By applying the network-based Man-In-The-Loop method on top of this, the accuracy of target identification is enhanced, and tracking performance of the target can be guaranteed through the terminal phase.

Example of Target Tracking Method (Source: Agency for Defense Development)

▲ Network-Based Aerial Platform Mission Control, The Future = Unlike previous battlefield environments where operator intervention was frequent, it is expected that operator involvement will be minimized in future battlefield environments. Moreover, as all systems become networked, the speed of information collection will accelerate, greatly enhancing battlefield responsiveness.

For all aerial platforms equipped with video sensors, applying a network environment enables accurate command and control of mission targets for multiple high-speed aerial platforms by utilizing video and spatial information on the ground with various situational awareness image processing technologies such as big data and deep learning.

Currently, due to infrastructure limitations, there are physical and technical constraints on high-speed aerial platforms transmitting high-resolution video. However, the Agency for Defense Development is continuously pursuing ways to overcome these environmental limitations by integrating technologies in remote control fields such as high-speed aerial platform video analysis technology, mission control technology, and integrated network system development technology to maximize battlefield responsiveness. Additionally, as technological maturity improves, these technologies are expected to be applicable in civilian sectors such as maritime control systems, maritime object targeting, and maritime object video information data generation.

This content was produced with the assistance of AI translation services.