As AI transformation expands in manufacturing, intelligent technologies are increasingly applied to CNC machine tools and machining processes. In multi-product, small-batch production environments, frequent product changes require flexible and autonomous process planning. This study proposes a standard data integration-based intelligent process planning system that automatically performs the entire process from 3D model input to NC code generation. To enable intelligent process planning, data across all stages—from feature recognition to machining execution—must be integrated into a unified flow and connected with AI-based decision-making. The proposed system uses an ISO 14649-based XML schema to sequentially link data generated by each module, ensuring standardized information flow. Based on this framework, rulebased feature recognition, constraint-based process planning, and machine learning-based cutting condition optimization are implemented. A prototype system was developed to validate the approach, automatically generating NC code for industrial parts and performing actual CNC machining. Experimental results confirmed the feasibility and validity of the proposed system. This study demonstrates that standardized data integration combined with AI technologies can enable autonomous, flexible, and efficient process planning for advanced manufacturing environments.
Recent manufacturing environments demand greater flexibility due to the increasing need for high-mix, low-volume production. While mobile and collaborative robots have made it easier to relocate equipment and change layouts, reconfiguring manufacturing cells remains challenging. Successful reconfiguration relies not only on physical layout changes but also on a deep understanding of the original design intent, operational constraints, and the empirical knowledge gained during operation. Unfortunately, this knowledge is often implicit and may depend on engineers or operators who are no longer available. To tackle this issue, this study introduces a framework for manufacturing cell reconfiguration based on the Asset Administration Shell (AAS). This framework integrates static engineering information with the operational knowledge acquired throughout construction and operation. It organizes asset specifications, operational states, manufacturing skills, and related documents into a unified structure, enabling reconfiguration decisions to reflect both system configurations and proven operating conditions. Furthermore, it connects work execution results with operational knowledge, document versions, and raw data references to enhance traceability and reproducibility post-reconfiguration. This proposed approach aims to reduce the complexity and cost of cell reconfiguration and relocation while enhancing operational flexibility, consistency, and scalability.
As smart factories evolve, maintenance manuals need to be transformed from static documents into machine-readable and reusable digital assets. However, many legacy manuals are still in unstructured formats, such as Hangul word-processor files, which complicates their updating, reusability, and adaptability to changing product configurations. This paper presents a framework for converting these legacy manuals into S1000D-based documents. It combines style-based hierarchy extraction with rule-guided multi-step transformation using a local large language model (LLM). First, the style information within the Korean documents is analyzed to identify the hierarchical structure of the manual and extract content at various document levels. Next, this extracted content is converted into S1000D XML modules through the local LLM, utilizing category-specific rule files, XML tag definitions, and example templates. To enhance structural consistency and minimize errors, different prompts and rule sets are applied based on the document hierarchy level.A case study involving a maintenance manual for a high-angle limit switch module demonstrates that the proposed method can maintain document structure while generating reusable S1000D-style outputs from legacy technical documents. This approach lays a practical foundation for creating continuously updatable and context-reconfigurable maintenance guidance in smart manufacturing environments.
This study introduces a wire-spring based planar gravity compensation mechanism and evaluates its performance through both analysis and experiments. The mechanism features three pulleys, one spring, and one wire, all arranged in a planar configuration for compact installation within a robotic arm. A linear approximation of the target gravitational torque was derived using the least-squares method, allowing for the determination of spring stiffness and initial tension. Experimental results indicated that the proposed mechanism reduced the maximum torque by approximately 63%. However, the measured slope was gentler than the theoretical model due to friction losses. Additional tests that varied spring stiffness (k) and initial wire tension (A) confirmed that k primarily influences the slope of the compensation torque, while A affects its intercept. This finding suggests that compensation performance can be tailored to specific requirements by adjusting these parameters. The study successfully demonstrates a compact and lightweight mechanism and experimentally validates its tunability through design adjustments. Future research will focus on reducing friction, extending the mechanism to multi-degree-of-freedom systems, and validating performance under dynamic conditions for applications in collaborative and medical robots.
Despite the increasing focus on the mental health of older adults and active senior populations, assessment tools still lag behind those for physical health monitoring. To bridge this gap, this study introduces an AI chatbot-based multimodal stress monitoring system that utilizes emotion recognition and heart rate variability (HRV). The system analyzes chatbot conversations, video, audio, and heart rate signals to assess facial expressions, speech emotions, and HRV, allowing for stress evaluation and user stratification into risk groups. Negative emotions are quantified and combined with HRV data to generate a stress score. Facial and speech emotion models were trained on the RAVDESS, CREMA, and TESS datasets, yielding 21,000 augmented samples through a BiLSTM network. Additionally, a deep learning-based HRV model utilized data from smartwatches to predict stress levels. By integrating facial, vocal, and HRV features through weighted fusion, the system produces a comprehensive stress index that categorizes users Healthy, Caution, Risk. This approach facilitates continuous monitoring at home, supporting early detection for preventive care and informed clinical decision-making.
Accurate localization in industrial environments is challenging due to factors such as dust and reflections that degrade perception. To overcome these limitations, we propose an environment-independent localization method that relies solely on ultra-wideband (UWB) positioning. Our system employs LiDAR-SLAM in an offline stage to create a global map frame and calibrate the transformation between this frame and the UWB anchors. During operation, the robot estimates its position using a Kalman filter applied to UWB measurements transformed into the map frame. This paper presents a preliminary feasibility study conducted in an office-like environment to verify the core calibration and localization pipeline. The results show that the proposed method effectively aligns UWB positions with a pre-built SLAM map, achieving a 94% reduction in root-mean-square error (RMSE) compared to raw UWB measurements when validated against LiDAR-SLAM ground truth. This initial verification establishes the technical viability of the framework and lays the groundwork for future validation in harsh, large-scale industrial settings.
Accurate 3D human pose reconstruction from a single RGB image remains challenging due to scale ambiguity and perspective distortions. Current single-view methods primarily rely on learned priors or kinematic constraints, but they often struggle to maintain geometric consistency with the physical scene. This results in horizon alignment drift and instability when rendered in metric environments. To overcome these limitations, this study introduces a vanishing-point-driven framework that integrates scene geometry into the pose correction process. Under the Manhattan-world assumption, dominant vanishing points are detected to estimate the ground plane and recover the camera orientation with high precision. A lightweight 3D pose estimation network generates initial joint coordinates in camera-centric space. These coordinates are then refined through a VP-based ground-alignment transformation, which resolves scale ambiguity and minimizes geometric drift. The corrected poses are normalized to physical scale and streamed to NVIDIA OmniverseTM for real-time digital-twin visualization. Experiments conducted on indoor scenes from the NYU Depth V2 dataset demonstrate sub-pixel accuracy in vanishing-point localization and significant improvements in geometric alignment between the reconstructed poses and the true scene layout. This confirms the effectiveness of the proposed approach for single-view digital-twin human modeling.
The automotive painting process is complex, featuring hybrid serial-parallel lines and unplanned repair operations, which makes production forecasting challenging. This study introduces an AI-driven predictive framework designed to estimate future work-in-process (WIP) in paint shops, with the goal of improving production management efficiency. We collected and preprocessed historical operational data through noise reduction and process filtering. Several machine learning and deep learning models were trained and validated. To ensure transparency, we utilized explainable AI (XAI) techniques. The proposed system proved feasible for deployment on a web-based monitoring platform, facilitating real-time decision-making in manufacturing environments.
This study presents an optimization framework for designing novel retainer rings (NRR) in chemical mechanical planarization (CMP) to enhance the uniformity of material removal rates (MRR). To improve optimization efficiency, we developed a finite element method (FEM) model alongside a Metamodel of Optimal Prognosis (MOP). The NRR outperformed the reference retainer ring (RRR) in our simulations. We classified simulation cases based on the pressure application area: long (LC), middle (MC), and short (SC). The MOP was constructed using Latin hypercube sampling and refined through an adaptive approach to achieve high accuracy while minimizing computational costs. Optimization was performed using an evolutionary algorithm, generating Pareto fronts for analysis. We evaluated representative designs based on MRR distribution and non-uniformity. Ultimately, Design 2-LC was identified as the optimal choice. The results indicate that the proposed framework effectively enhances MRR uniformity while reducing optimization time.
The formation of a hat-profile is significantly influenced by springback and the final cross-sectional geometry, both of which are sensitive to die profile design. This study introduces a scalar-based artificial neural network (ANN) surrogate model combined with genetic-algorithm (GA) optimization to enhance die and process design efficiency. An automated ABAQUS finite-element workflow was established to generate 900 design cases. For each case, seven scalar geometric and angle responses characterizing the post-forming cross section were extracted and used to train a multilayer perceptron. This network maps four die design variables to the final geometry. The surrogate model demonstrated high predictive accuracy, with geometric and angular errors remaining small and coefficients of determination (R2) nearing 1.0. This enabled quick evaluation of new designs without the need for additional finiteelement analyses. By integrating the ANN surrogate within a GA, optimal die geometries were identified that reduce springback while meeting target dimensions, showcasing the proposed framework as an effective AI-driven design tool for sheet-metal forming.
This study evaluates the load and moment characteristics of composite leaf springs used in the front suspension of a 4.0- ton gross vehicle weight (GVW) light commercial van through CarSim-based vehicle dynamics simulations. Carbon fiber composite (CFC), glass fiber composite (GFC), and hybrid composite (HC, carbon 20%: glass 80%) leaf springs were fabricated with identical geometry using a prepreg compression molding (PCM) process. Spring constants obtained from four-point bending tests were incorporated into the vehicle dynamics model. Dynamic responses were analyzed under flatroad driving, acceleration, braking, cornering, and speed bump conditions. The results indicate that the GFC leaf spring achieved a 61.5% weight reduction compared to a conventional steel spring while maintaining equivalent vertical load and roll moment responses. The HC exhibited improved roll suppression and pitch stability, whereas the CFC demonstrated excessively high stiffness, limiting its applicability to heavy-duty vehicles. Furthermore, the GFC maintained stable dynamic performance after low-velocity impact damage of 20 and 80 J, with stiffness remaining within ±5% of the steel reference. These findings confirm that composite leaf springs, particularly those made from glass fiber composites, provide a practical and durable alternative to steel leaf springs for light commercial vehicle suspension systems.
Dry electrode fabrication is considered a crucial next-generation process for secondary batteries because it eliminates the need for solvents and drying steps, significantly reducing energy consumption and carbon emissions. To achieve optimal performance in dry electrodes, it is essential to ensure high mechanical stability and electrical conductivity. These properties can be enhanced by controlling binder fibrillation and creating a continuous conductive network through the uniform dispersion of conductive additives. In this study, we applied mechanical shear mixing as a pre-treatment to electrode powders, which included active materials, conductive agents, and binders. We systematically investigated variations in electrical conductivity, binding structure, tensile properties, internal resistance (via IR drop), and fast-charging performance as a function of the mixing shear rate. In particular, we quantified the binder fibrillation behavior and the dispersion of conductive agents that occur simultaneously during mixing. By correlating these factors with the physical and electrochemical properties of the final electrode film, we propose design guidelines to optimize the mixing pre-treatment process.
Estimating lower-limb joint torques during gait using inertial measurement units (IMUs) has attracted growing attention in biomechanics and wearable sensing. Conventional approaches rely on inverse dynamics based on segmental kinematics and ground reaction forces, requiring force sensors or full-body sensor setups. This study proposes a recurrent neural network (RNN) method to estimate lower-limb joint torques using segmental kinematic data from a limited number of IMUs.Twelve healthy participants performed treadmill walking and running under twelve different conditions to generate training data. Model inputs included center-of-mass accelerations and angular velocities of the pelvis and shank.Results demonstrated two key findings. First, a model using three IMUs achieved performance comparable to a seven-IMU model, with hip flexion torque errors of approximately 0.18 Nm/kg, demonstrating strong effectiveness with a reduced sensor configuration. Second, while inverse dynamics exhibited an error increase of 0.28 Nm/kg from the ankle to the hip, the proposed model showed only a 0.01 Nm/kg increase and achieved approximately 0.13 Nm/kg lower error at the hip.These results indicate that accurate and efficient joint torque estimation is feasible using an RNN with fewer wearable sensors.
The nano satellite industry has transitioned to low-cost development, driven by private companies and research organizations in the NewSpace era. Can-Satellite offers a budget-friendly alternative to traditional cube satellite manufacturing and testing. This study focuses on enhancing the reliability of small satellite designs by analyzing the vibration stability of PLA plates, the primary structure of a Can-Satellite, produced through Fused Filament Fabrication (FFF) 3D printing. Quasi-static, modal, and random vibration analyses were conducted using Finite Element Analysis (FEA) with ANSYS to evaluate stacking directions along the x, y, and z axes and optimize structural stability. The findings indicate that the y-axis laminated structure exhibits superior vibration endurance, effectively reducing issues during launch. This research contributes to improving the reliability of Can-Satellites and enhances manufacturing efficiency for cube and micro-satellite projects. Additionally, it supports the advancement of educational satellites and domestic small satellite technology.
First
Prev
