#1. On Dec. 19 last year, at LIG Nex1's research and development (R&D) Center 2 Pangyo House in Bundang District, Seongnam, Gyeonggi. A notice appeared on the large screen in the command and control lab (LAB) reading, "War breaks out D+7, North Korean troops have occupied the Paju factory rent(expense), and additional forces are moving south." On the map set against the Paju area in Gyeonggi, the positions of North Korean tanks, armored vehicles, and troops were shown, and the locations of friendly forces by weapons systems, including drones and tanks, were also displayed in detail. The system classified the current risk level as "danger."
On another monitor, a program similar to the generative artificial intelligence (AI) ChatGPT provided video reading results along with the message, "Here is the requested situation summary." When the commander typed, "Based on the risk assessment, generate the strike scheduling results report," various figures on the large screen changed immediately. A threat priority list appeared from tanks to people, and it showed that the South Korean military was conducting counterfire using K9 self-propelled howitzers and Hyeongung, among others. As friendly forces began their attack, the number of enemy troops and the risk level decreased in real time.
#2. On Dec. 22 last year, at the production site of Korea Aviation Engineering and Maintenance Service (KAEMS), an affiliate of Korea Aerospace Industries (KAI), in Sacheon, South Gyeongsang. Passing the line where the forward fuselage of the FA-50 multirole fighter bound for Poland was being built, white-and-orange drones came into view. Five employees were inserting various components inside these drones, and engines to be mounted on the drones were placed on the floor. The aircraft is a multirole drone (AAP) that KAI is developing.
As the importance of "sovereign AI (AI sovereignty)" has recently come to the fore in the global defense industry market, the domestic defense and defense industry sectors are rapidly pushing ahead with the introduction of AI-powered weapons systems and the development of unmanned weapons. Sovereign AI refers to AI systems that allow each country to hold and control its own data without relying on external clouds or services.
Domestic defense companies focused through last year on building the framework to apply AI, but starting this year they plan to begin demonstrating and advancing it. We looked into the sites where LIG Nex1 and KAI are pioneering the future of Korea's weapons manufacturing industry.
◇ Reducing decision time in emergencies… LIG Nex1 develops AI battle staff
The command and control system (C4I) under development by LIG Nex1 refers to a command and control framework that integrates the military's command, control, communication, computer and intelligence functions.
The AI C4I under development is characterized by leveraging LLM-based AI. It analyzes and predicts information secured by reconnaissance assets in real time, communicates with the commander in real time, and shortens the speed of operational decision-making on the battlefield.
LIG Nex1 applied the "sensor-to-shooter" concept to this system, enabling immediate sharing with C4I of information derived on the battlefield, friendly force capabilities, and other domains, all the way through to response. LIG Nex1's plan is to support faster information acquisition and decision-making by commanders than currently fielded C4I.
A LIG Nex1 official said, "The core of this system is multiple AIs that analyze each type of information and interlock in a complex way," adding, "We aim for precise analysis of the battlefield environment and effective operational responses."
LIG Nex1 calls the technology embedded in this C4I "AI automatic collaboration technology." Unlike existing technology where a single AI model handled everything from battlefield situation analysis to response, this technology makes multiple AI models collaborate.
This means each AI computes and derives the optimal conclusion, as if operations staff and tactics staff gathered for a meeting. LIG Nex1 is currently working with AI corporations to train the AI, and this year plans to continuously train it on defense data to make it "defense-specialized."
Along with AI advancement, research is underway to mount C4I on a mobile platform. The goal is to build a Korean version of Titan by U.S. AI company Palantir. Titan is a combat command vehicle that collects and analyzes combat-related data through various sensors and provides strategic information to soldiers in real time.
◇ KAI links AAP drone with FA-50… completing a manned-unmanned teaming system
KAI is currently producing eight AAPs. A total of 10 will be built, and two—AAP-001 and 002—are already complete and undergoing test flights. The AAP consists of a nose where mission equipment is mounted; an upper fuselage where a computer to aid AI flight is installed; a lower fuselage with a parachute between two fuel tanks; and a tail where the engine will be attached. If a warhead is mounted as mission equipment, it becomes a loitering munition; if a camera or radar is attached, it becomes a reconnaissance drone.
What stood out at KAI's AAP production site were the AAP-009 and 010 aircraft, designed in blue on a gray base. They were built to link with the FA-50 fighter jointly developed by KAI and Lockheed Martin of the United States. They will be投入 for manned-unmanned teaming demonstrations. KAI plans to mount an AI pilot on these aircraft to complete the manned-unmanned teaming system.
All processes are being carried out with KAI's own investment. More than 1 billion won in expense has been投入 to build a single AAP, but once development is completed and it moves to mass production, the unit production cost is expected to fall below 100 million won.
Alongside aircraft production, work is also underway to advance "Kailot," KAI's AI pilot. Through the first half of last year, KAI produced a scaled-down AAP and conducted test flights. Kang Byung-gil, head of KAI's future systems research office, said, "A test aircraft applying Kailot avoided multiple fixed and moving obstacles installed along the flight path and reached the target point."
KAI's goal is to complete the linkage with the FA-50 by 2027. Demonstrations will proceed with the front-seat pilot flying the FA-50 and the back-seat pilot managing the drone. Kang said, "We plan to link the FA-50 and AAP via a data link and apply AI algorithms for demonstrations."