%53%65%75%6e%67%68%77%61%20%52%79%75
-
KAIST, AI judges manufacturing beyond craftsmanship and language barriers
<(From Left) M.S candidate Inhyo Lee, Ph.D candidate Heekyu Kim, Ph.D candidate joonyoung Kim, Professor Seunghwa Ryu>
Most of the plastic products we use are made through injection molding, a process in which molten plastic is injected into a mold to mass-produce identical items. However, even slight changes in conditions can lead to defects, so the process has long relied on the intuition of highly skilled workers. Now, KAIST researchers have proposed an AI-based solution that autonomously optimizes processes and transfers manufacturing knowledge, addressing concerns that expertise could be lost due to the retirement of skilled workers and the increase in foreign labor.
KAIST (President Kwang Hyung Lee) announced on the 22nd of December that a research team led by Professor Seunghwa Ryu from the Department of Mechanical Engineering · InnoCORE PRISM-AI Center has, for the first time in the world, developed generative AI technology that autonomously optimizes injection molding processes, along with an LLM-based knowledge transfer system that makes on-site expertise accessible to anyone. The team also reported that these achievements were published consecutively in an internationally renowned journal.
The first achievement is a generative AI–based process inference technology that automatically infers optimal process conditions based on environmental changes or quality requirements. Previously, whenever temperature, humidity, or desired quality levels changed, skilled workers had to rely on trial and error to readjust conditions.
The research team implemented a diffusion model–based approach that reverse-engineers process conditions satisfying target quality requirements, using environmental data and process parameters collected over several months from an actual injection molding factory.
In addition, the team built a surrogate model that substitutes for actual production, enabling quality prediction without running the real process. As a result, they achieved an error rate of just 1.63%, significantly lower than the 23~44% error rates of representative existing technologies such as GAN* and VAE** models traditionally used for process prediction. Experiments applying the AI-generated conditions to real processes confirmed successful production of acceptable products, demonstrating practical applicability.
*GAN (Generative Adversarial Network): a method in which two AI models compete with each other to generate data
**VAE (Variational Autoencoder): a method that compresses and learns common patterns in data and then reconstructs them
<Figure 1. Generative AI–Based Process Reasoning Technology>
The second achievement is the IM-Chat, an LLM-based knowledge transfer system designed to address skilled worker retirement and multilingual work environments. IM-Chat is a multi-agent AI system that combines large language models (LLMs) with retrieval-augmented generation (RAG), serving as an AI assistant for manufacturing sites by providing appropriate solutions to problems encountered by novice or foreign workers.
When a worker asks a question in natural language, the AI understands it and, if necessary, automatically calls the generative process inference AI, simultaneously providing optimal process condition calculations along with relevant standards and background explanations.
For example, when asked, “What is the appropriate injection pressure when the factory humidity is 43.5%?”, the AI calculates the optimal condition and presents the supporting manual references as well. With support for multilingual interfaces, foreign workers can receive the same level of decision-making support.
This research is regarded as a core manufacturing AI transformation (AX) technology that can be extended beyond injection molding to molds, presses, extrusion, 3D printing, batteries, bio-manufacturing, and other industries.
In particular, the work is significant in that it presents a paradigm for autonomous manufacturing AI, integrating generative AI and LLM agents through a Tool-Calling approach*, enabling AI to make its own judgments and invoke necessary functions.
*Tool-Calling approach: a method in which AI autonomously calls and uses the functions or programs required for a given situation
<Figure 2. Large Language Model–Based Multilingual Knowledge Transfer Multi-Agent IM-Chat>
<Figure 3. Example of Operation of the Large Language Model (LLM)–Based Multilingual Knowledge Transfer Multi-Agent IM-Chat>
<Figure 4. Illustration of the Application of an LLM-Based Multilingual Knowledge Transfer Multi-Agent IM-Chat (AI-Generated)>
Professor Seunghwa Ryu explained, “This is a case where we addressed fundamental problems in manufacturing in a data-driven way by combining AI that autonomously optimizes processes with LLMs that make on-site knowledge accessible to anyone,” adding, “We will continue expanding this approach to various manufacturing processes to accelerate intelligence and autonomy across the industry.”
This research involved doctoral candidates Junhyeong Lee, Joon-Young Kim, and Heekyu Kim from the Department of Mechanical Engineering as co–first authors, with Professor Seunghwa Ryu as the corresponding author. The results were published consecutively in the April and December issues of Journal of Manufacturing Systems (JCR 1/69, IF 14.2), the world’s top-ranked international journal in engineering and industrial fields.
※ Paper 1: “Development of an Injection Molding Production Condition Inference System Based on Diffusion Model,” DOI: https://doi.org/10.1016/j.jmsy.2025.01.008 ※ Paper 2: “IM-Chat: A multi-agent LLM framework integrating tool-calling and diffusion modeling for knowledge transfer in injection molding industry,” DOI: https://doi.org/10.1016/j.jmsy.2025.11.007
This research was supported by the Ministry of Science and ICT, the Ministry of SMEs and Startups, and the Ministry of Trade, Industry and Energy.
2025.12.22 View 3293 -
Physics Informed AI Excels at Large Scale Discovery of New Materials!
<(From left) Ph.D candidates Songho Lee, Donggeun Park, and Hyeonbin Moon, and Professor Seunghwa Ryu from the Department of Mechanical Engineering; (top) Professor Jae Hyuk Lim from Kyung Hee University and Dr. Wabi Demeke from KAIST>
One of the key steps in developing new materials is “property identification,” which has long relied on massive amounts of experimental data and expensive equipment, limiting research efficiency. A KAIST research team has introduced a new technique that combines “physical laws,” which govern deformation and interaction of materials and energy, with artificial intelligence. This approach allows for rapid exploration of new materials even under data-scarce conditions and provides a foundation for accelerating design and verification across multiple engineering fields, including materials, mechanics, energy, and electronics.
KAIST (President Kwang Hyung Lee) announced on the 2nd of October that Professor Seunghwa Ryu’s research group in the Department of Mechanical Engineering, in collaboration with Professor Jae Hyuk Lim’s group at Kyung Hee University (President Jinsang Kim) and Dr. Byungki Ryu at the Korea Electrotechnology Research Institute (President Namkyun Kim), proposed a new method that can accurately determine material properties with only limited data. The method uses Physics-Informed Machine Learning (PIML), which directly incorporates physical laws into the AI learning process.
<Schematic Diagram of a Physics-Based Machine Learning Methodology for Understanding Material Properties>
In the first study, the researchers focused on hyperelastic materials, such as rubber. They presented a Physics-Informed Neural Network (PINN) method that can identify both the deformation behavior and the properties of materials using only a small amount of data obtained from a single experiment. Whereas previous approaches required large, complex datasets, this research demonstrated that material characteristics can be reliably reproduced even when data is scarce, limited, or noisy.
In the second study, the team turned to thermoelectric materials—new materials that convert heat into electricity and electricity into heat. They proposed a PINN-based inverse inference technique that can estimate key indicators, such as thermal conductivity (how well heat is transferred) and the Seebeck coefficient (how efficiently electricity is generated), from just a few measurements.
Going further, the researchers introduced a Physics-Informed Neural Operator (PINO), an AI model that understands the physical laws of nature, and showed that it can generalize to previously unseen materials without requiring retraining.
In fact, after training the system on 20 materials, they tested it on 60 entirely new materials, and in all cases it predicted their properties with high accuracy. This breakthrough points to a future where large-scale, high-speed screening of countless candidate materials becomes possible.
This achievement goes beyond simply reducing the need for experiments. By intricately combining physical laws with AI, the researchers provided the first example of improving experimental efficiency while preserving reliability.
Professor Seunghwa Ryu, who led both studies, stated, “This is the first case of applying AI that understands physical laws to real material research. It enables reliable identification of material properties even when data availability is limited, and it is expected to expand into various engineering fields.”
The first paper, co-first-authored by KAIST Mechanical Engineering PhD candidates Hyeonbin Moon and Donggeun Park, was published on August 13 in Computer Methods in Applied Mechanics and Engineering.
※ Paper title: “Physics-informed neural network-based discovery of hyperelastic constitutive models from extremely scarce data”
※ DOI: https://doi.org/10.1016/j.cma.2025.118258
The second paper, co-first-authored by KAIST Mechanical Engineering PhD candidates Hyeonbin Moon and Songho Lee, and Dr. Wabi Demeke, was published on August 22 in npj Computational Materials.
※ Paper title: “Physics-informed neural operators for generalizable and label-free inference of temperature-dependent thermoelectric properties” ※ DOI: https://doi.org/10.1038/s41524-025-01769-1
Meanwhile, the first study was supported by the Korea Research Foundation and the Ministry of Science and ICT’s INNOCore Program, as well as by a research project from the Ministry of Food and Drug Safety. The second study was carried out with support from the Korea Research Foundation and the Ministry of Science and ICT’s INNOCore Program.
2025.10.10 View 3769 -
Ultralight advanced material developed by KAIST and U of Toronto
< (From left) Professor Seunghwa Ryu of KAIST Department of Mechanical Engineering, Professor Tobin Filleter of the University of Toronto, Dr. Jinwook Yeo of KAIST, and Dr. Peter Serles of the University of Toronto >
Recently, in advanced industries such as automobiles, aerospace, and mobility, there has been increasing demand for materials that achieve weight reduction while maintaining excellent mechanical properties. An international joint research team has developed an ultralight, high-strength material utilizing nanostructures, presenting the potential for various industrial applications through customized design in the future.
KAIST (represented by President Kwang Hyung Lee) announced on the 18th of February that a research team led by Professor Seunghwa Ryu from the Department of Mechanical Engineering, in collaboration with Professor Tobin Filleter from the University of Toronto, has developed a nano-lattice structure that maximizes lightweight properties while maintaining high stiffness and strength.
In this study, the research team optimized the beam shape of the lattice structure to maintain its lightweight characteristics while maximizing stiffness and strength.
Particularly, using a multi-objective Bayesian optimization algorithm*, the team conducted an optimal design process that simultaneously considers tensile and shear stiffness improvement and weight reduction. They demonstrated that the optimal lattice structure could be predicted and designed with significantly less data (about 400 data points) compared to conventional methods.
*Multi-objective Bayesian optimization algorithm: A method that finds the optimal solution while considering multiple objectives simultaneously. It efficiently collects data and predicts results even under conditions of uncertainty.
< Figure 1. Multi-objective Bayesian optimization for generative design of carbon nanolattices with high compressive stiffness and strength at low density. The upper is the illustration of process workflow. The lower part shows top four MBO CFCC geometries with their 2D Bézier curves. (The optimized structure is predicted and designed with much less data (approximately 400) than the conventional method >
Furthermore, to maximize the effect where mechanical properties improve as size decreases at the nanoscale, the research team utilized pyrolytic carbon* material to implement an ultralight, high-strength, high-stiffness nano-lattice structure.
*Pyrolytic carbon: A carbon material obtained by decomposing organic substances at high temperatures. It has excellent heat resistance and strength, making it widely used in industries such as semiconductor equipment coatings and artificial joint coatings, where it must withstand high temperatures without deformation.
For this, the team applied two-photon polymerization (2PP) technology* to precisely fabricate complex nano-lattice structures, and mechanical performance evaluations confirmed that the developed structure simultaneously possesses strength comparable to steel and the lightness of Styrofoam.
*Two-photon polymerization (2PP) technology: An advanced optical manufacturing technique based on the principle that polymerization occurs only when two photons of a specific wavelength are absorbed simultaneously.
Additionally, the research team demonstrated that multi-focus two-photon polymerization (multi-focus 2PP) technology enables the fabrication of millimeter-scale structures while maintaining nanoscale precision.
Professor Seunghwa Ryu explained, "This technology innovatively solves the stress concentration issue, which has been a limitation of conventional design methods, through three-dimensional nano-lattice structures, achieving both ultralight weight and high strength in material development."
< Figure 2. FESEM image of the fabricated nano-lattice structure and (bottom right) the macroscopic nanolattice resting on a bubble >
He further emphasized, "By integrating data-driven optimal design with precision 3D printing technology, this development not only meets the demand for lightweight materials in the aerospace and automotive industries but also opens possibilities for various industrial applications through customized design."
This study was led by Dr. Peter Serles of the Department of Mechanical & Industrial Engineering at University of Toronto and Dr. Jinwook Yeo from KAIST as co-first authors, with Professor Seunghwa Ryu and Professor Tobin Filleter as corresponding authors.
The research was published on January 23, 2025 in the international journal Advanced Materials (Paper title: “Ultrahigh Specific Strength by Bayesian Optimization of Lightweight Carbon Nanolattices”).
DOI: https://doi.org/10.1002/adma.202410651
This research was supported by the Multiphase Materials Innovation Manufacturing Research Center (an ERC program) funded by the Ministry of Science and ICT, the M3DT (Medical Device Digital Development Tool) project funded by the Ministry of Food and Drug Safety, and the KAIST International Collaboration Program.
2025.02.18 View 13408 -
Deep Learning Framework to Enable Material Design in Unseen Domain
Researchers propose a deep neural network-based forward design space exploration using active transfer learning and data augmentation
A new study proposed a deep neural network-based forward design approach that enables an efficient search for superior materials far beyond the domain of the initial training set. This approach compensates for the weak predictive power of neural networks on an unseen domain through gradual updates of the neural network with active transfer learning and data augmentation methods.
Professor Seungwha Ryu believes that this study will help address a variety of optimization problems that have an astronomical number of possible design configurations. For the grid composite optimization problem, the proposed framework was able to provide excellent designs close to the global optima, even with the addition of a very small dataset corresponding to less than 0.5% of the initial training data-set size. This study was reported in npj Computational Materials last month.
“We wanted to mitigate the limitation of the neural network, weak predictive power beyond the training set domain for the material or structure design,” said Professor Ryu from the Department of Mechanical Engineering.
Neural network-based generative models have been actively investigated as an inverse design method for finding novel materials in a vast design space. However, the applicability of conventional generative models is limited because they cannot access data outside the range of training sets. Advanced generative models that were devised to overcome this limitation also suffer from weak predictive power for the unseen domain.
Professor Ryu’s team, in collaboration with researchers from Professor Grace Gu’s group at UC Berkeley, devised a design method that simultaneously expands the domain using the strong predictive power of a deep neural network and searches for the optimal design by repetitively performing three key steps.
First, it searches for few candidates with improved properties located close to the training set via genetic algorithms, by mixing superior designs within the training set. Then, it checks to see if the candidates really have improved properties, and expands the training set by duplicating the validated designs via a data augmentation method. Finally, they can expand the reliable prediction domain by updating the neural network with the new superior designs via transfer learning. Because the expansion proceeds along relatively narrow but correct routes toward the optimal design (depicted in the schematic of Fig. 1), the framework enables an efficient search.
As a data-hungry method, a deep neural network model tends to have reliable predictive power only within and near the domain of the training set. When the optimal configuration of materials and structures lies far beyond the initial training set, which frequently is the case, neural network-based design methods suffer from weak predictive power and become inefficient.
Researchers expect that the framework will be applicable for a wide range of optimization problems in other science and engineering disciplines with astronomically large design space, because it provides an efficient way of gradually expanding the reliable prediction domain toward the target design while avoiding the risk of being stuck in local minima. Especially, being a less-data-hungry method, design problems in which data generation is time-consuming and expensive will benefit most from this new framework.
The research team is currently applying the optimization framework for the design task of metamaterial structures, segmented thermoelectric generators, and optimal sensor distributions. “From these sets of on-going studies, we expect to better recognize the pros and cons, and the potential of the suggested algorithm. Ultimately, we want to devise more efficient machine learning-based design approaches,” explained Professor Ryu.This study was funded by the National Research Foundation of Korea and the KAIST Global Singularity Research Project.
-Publication
Yongtae Kim, Youngsoo, Charles Yang, Kundo Park, Grace X. Gu, and Seunghwa Ryu, “Deep learning framework for material design space exploration using active transfer learning and data augmentation,” npj Computational Materials (https://doi.org/10.1038/s41524-021-00609-2)
-Profile
Professor Seunghwa Ryu
Mechanics & Materials Modeling Lab
Department of Mechanical Engineering
KAIST
2021.09.29 View 20249