This study proposes command control system that enables military operators to intuitively control multiple Unmanned Vehicles(UVs) using natural language commands. The system employs a Sequence-to-Sequence (Seq2Seq)-based Small Language Model(SLM) that is pre-trained to interpret natural language and convert it into executable UV commands. Operators issue commands via a microphone, and the system analyzes the input to assign appropriate tasks to each UV while reporting execution results in real time. This approach reduces the training burden on military personnel, enables efficient control of multiple UVs, and enhances transparency in the command-response structure. The proposed system provides a practical technical foundation for improving the operational efficiency and reliability of UV operations in real combat environments.

To carry out missions effectively in battlefield environments, it is essential to rapidly identify enemy threats and respond through precise analysis. Accordingly, technologies that utilize artificial intelligence to identify and classify targets in real time are being actively researched in modern warfare. In this study, we propose a deep learning-based target classification algorithm that simultaneously ensures real-time performance and high classification accuracy. Time-series data are constructed from infrared imagery and augmented to train the model, and the effectiveness of the proposed algorithm is demonstrated through comparative experiments with various CNN- and RNN-based models.

Modern person detection models are typically trained on urban-centric datasets where targets are visually distinct from their surroundings. However, in forested environments, military uniforms and complex natural backgrounds significantly degrade detection performance. To address this gap, we collect a dataset of individuals wearing military uniforms in wooded areas, simulating realistic low-altitude search and surveillance scenarios. We demonstrate that state-of-the-art YOLO series detectors, widely used for real-time inference, show substantial performance drops when applied to these challenging settings. Using our dataset, we train and evaluate YOLObased models to quantify their effectiveness in camouflage scenarios. Our findings highlight the limitations of existing detectors in recognizing individuals in military uniforms and provide a foundation for developing models tailored to such environments.

Guided missiles can be loaded in canisters to protect from external environments. Accordingly, guided missiles should be rigidly fixed to the canisters to endure vibration and shock conditions. On the other hand, guided missiles should be rapidly released when it is launched. separation bolts are developed and implemented to meet those conflicting requirements. However, pyroshock generated by separation bolts could cause lethal damage to other devices installed on guided missiles. In this paper, shock levels are measured and classified by operating test of actual guided missiles and canister assembly. As a result, pyroshock test specifications are suggested in terms of distance from separation bolts. The results are expected to be applied to design and test for survivability assessment of guided missile equipment.

This study presents a method to evaluate and fuse detection information for surface-to-air missile (SAM) threats based on Bayesian search theory and radar equations. Detection probabilities are computed using radar parameters, and spatial distributions are visualized through MATLAB simulations based on two radar warning receiver models. Results indicate that detection outcomes vary depending on threat type and receiver location. Mid-range SAMs show stable detection patterns, while long-range SAM presents low detection probability, limiting its geolocation accuracy. Bayesian updating is applied to integrate multi-source detection results, resulting in improved posterior localization for specific categories of surface-to-air missile systems distinguished by range and detection characteristics. This approach provides a quantitative framework for integrating surface-to-air threat information in operational environments and contributes to the development of future electronic warfare and air defense strategies.

In conventional air-gapped control central console, user manuals have been provided in PDF format to support user comprehension of operational concepts and maintenance procedures. However, due to the extensive length of these documents, often comprising several hundred pages, users face significant difficulty in locating relevant information efficiently. To address this issue, this study proposes an on-device AI chatbot system that enables user manual-based question answering by leveraging lightweight NLP and LLM models suitable for deployment in air-gapped environments. For the performance evaluation of the system, response accuracy and the relevance of the associated page information were measured, confirming a high level of question-answering quality.

The recent war patterns, such as the Russia-Ukraine war, show that it is becoming important to integrate and operate various unmanned platforms for complex tactical operations. Therefore, cooperative operability between various platforms and an integrated control system using various controllers are becoming increasingly essential elements in the modern battlefield. In such an integrated control system, the operator's operational convenience must be considered to improve the task performance capability. Based on HCI technology, the way to promote high trust of operators is to increase the familiarity of the robot. Also, unmanned weapon systems generally develop GCS(Ground Control Station) and portable controller together for emergency response, but the functionality of portable controller is limited. In this paper, we propose a SW-common heterogeneous controller through a familiar multiple control input devices based on HCI. This paper measured the performance and reliability of remote control technology in simulated environment using TiA(Trust in Automation), an autonomous/semi-autonomous feeling performance indicator for heterogeneous multiple controllers, and NASA-TLX(NASA-Task Load Index), a workload evaluation tool. We propose a highly reliable multi Unmanned-platform remote control technology proven through the corresponding indicators.

This paper investigates the effect of temperature variation on the performance of a gimbal-servo system. The friction torque of the system was characterized through hysteresis loop measurements under different temperature conditions. Additionally, frequency response analysis was conducted, focusing on key results such as peak gain, gain margin, and phase margin. Experimental findings indicate that the Coulomb friction torque increases as the temperature decreases. Moreover, in the frequency response analysis, a rise in the damping coefficient was observed in the low-frequency region at lower temperatures, whereas the high-frequency region exhibited minimal temperature-dependent variation. To explain this behavior, this study models the Coulomb friction torque using an equivalent viscous damping coefficient that is inversely proportional to frequency.