Yangjin Kim
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Yangjin Kim has a total of 65 co-authors. Here are the top co-authors by number of shared publications:
Naohiko Sugita
Department of Mechanical Engineering, The University of Tokyo, Tokyo, Japan
20 shared publicationsSungtae Kim
DN Solutions, Gyeongsangnam-do 51537, Republic of Korea
12 shared publicationsYusuke Ito
Department of Mechanical Engineering, The University of Tokyo, Tokyo, Japan
11 shared publicationsKenichi Hibino
10 shared publicationsMamoru Mitsuishi
9 shared publicationsJurim Jeon
7 shared publications
Latest Publications
- Open Access
- 1 July 2026
- Published by International Journal of Precision Engineering and Manufacturing-Smart Technology of Korean Society for Precision Engineering in International Journal of Precision Engineering and Manufacturing-Smart Technology
- Vol. 4 (2) , 177-187
Variance-based Sensitivity Analysis for Efficient Volumetric Error Compensation in 5-axis Machine Tools
Conventional volumetric error compensation for 5-axis machine tools considers all kinematic errors to maximize machining accuracy, resulting in high computational complexity. This study proposes an efficient error compensation algorithm based on variance-based sensitivity analysis. Since kinematic errors are the primary source of volumetric errors, their relative influence was quantified to identify parameters with negligible effects. A volumetric error model incorporating 11 kinematic errors and three setup errors was developed using homogeneous transformation matrices (HTMs). Variance-based sensitivity analysis combined with Quasi-Monte Carlo (QMC) sampling was then applied to determine which kinematic errors could be excluded from the compensation process. Rather than adopting a fixed parameter reduction, the proposed method provides a flexible strategy for selecting compensation parameters according to sensitivity. Experimental results showed that excluding a limited number of non-dominant parameters maintained compensation accuracy while improving parameter identifiability and reducing model complexity. The reduced models also lowered the computational cost of parameter estimation by 19.9–37.0% compared with the full model. These results demonstrate that the proposed framework provides an effective and practical approach for optimizing volumetric error compensation in 5-axis machine tools.
- 1 July 2026
- Published by International Journal of Precision Engineering and Manufacturing-Smart Technology of Korean Society for Precision Engineering in International Journal of Precision Engineering and Manufacturing-Smart Technology
- Vol. 4 (2) , 177-187
Open AccessVariance-based Sensitivity Analysis for Efficient Volumetric Error Compensation in 5-axis Machine Tools
Conventional volumetric error compensation for 5-axis machine tools considers all kinematic errors to maximize machining accuracy, resulting in high computational complexity. This study proposes an efficient error compensation algorithm based on variance-based sensitivity analysis. Since kinematic errors are the primary source of volumetric errors, their relative influence was quantified to identify parameters with negligible effects. A volumetric error model incorporating 11 kinematic errors and three setup errors was developed using homogeneous transformation matrices (HTMs). Variance-based sensitivity analysis combined with Quasi-Monte Carlo (QMC) sampling was then applied to determine which kinematic errors could be excluded from the compensation process. Rather than adopting a fixed parameter reduction, the proposed method provides a flexible strategy for selecting compensation parameters according to sensitivity. Experimental results showed that excluding a limited number of non-dominant parameters maintained compensation accuracy while improving parameter identifiability and reducing model complexity. The reduced models also lowered the computational cost of parameter estimation by 19.9–37.0% compared with the full model. These results demonstrate that the proposed framework provides an effective and practical approach for optimizing volumetric error compensation in 5-axis machine tools.
- Open Access
- 9 May 2026
- Published by Elsevier in International Journal of Machine Tools and Manufacture
- Vol. 218
Adaptive and deterministic laser polishing enabled by in-process interferometric feedback
- 9 May 2026
- Published by Elsevier in International Journal of Machine Tools and Manufacture
- Vol. 218
Open AccessAdaptive and deterministic laser polishing enabled by in-process interferometric feedback
- Open Access
- 28 April 2026
- Published by Elsevier in CIRP Annals
- Vol. 75 (1) , 251-255
Closed-loop laser planarization of diamond guided by white-light interferometry
- 28 April 2026
- Published by Elsevier in CIRP Annals
- Vol. 75 (1) , 251-255
Open AccessClosed-loop laser planarization of diamond guided by white-light interferometry
- 27 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 75 (00189456) , 1-13
Simultaneous Surface and Thickness Measurement of Transparent Substrate Based on Physics-Informed Deep Learning
Transparent optical substrates such as photomask blanks and glass plates require precise metrology of the surface shape and thickness variation to ensure dimensional integrity in the ultra-precision industry. While conventional wavelength-tuning Fizeau interferometry can measure these properties simultaneously, it suffers from long acquisition times and systematic error accumulation owing to its reliance on numerous interferograms. To eliminate these limitations, we propose a highly efficient simultaneous measurement framework using physics-informed deep learning, which can extract three phase distributions using only two arbitrarily phase-shifted interferograms. We develop two deep learning-based modules: Pix2Fringe, which decomposes superimposed interferograms into independent fringe patterns using a pairwise cross-loss and refined phase relation to maintain physical consistency, and Pix2Phase, which reconstructs phase maps corresponding to the top surface, bottom surface, and total thickness variation. Experimental validation with a Fizeau interferometer demonstrates performance comparable to conventional approaches using up to 37 interferograms, while reducing systematic error accumulation and execution time by more than 22 times. Physical-constraint evaluation further confirms stable and physically consistent reconstruction, with only random residual fluctuations and a root-mean-square error of 0.0017 μm. These results highlight the potential of the proposed framework for efficient interferometric metrology in dynamic applications such as in-line and in-process inspection.
- 27 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 75 (00189456) , 1-13
Simultaneous Surface and Thickness Measurement of Transparent Substrate Based on Physics-Informed Deep Learning
Transparent optical substrates such as photomask blanks and glass plates require precise metrology of the surface shape and thickness variation to ensure dimensional integrity in the ultra-precision industry. While conventional wavelength-tuning Fizeau interferometry can measure these properties simultaneously, it suffers from long acquisition times and systematic error accumulation owing to its reliance on numerous interferograms. To eliminate these limitations, we propose a highly efficient simultaneous measurement framework using physics-informed deep learning, which can extract three phase distributions using only two arbitrarily phase-shifted interferograms. We develop two deep learning-based modules: Pix2Fringe, which decomposes superimposed interferograms into independent fringe patterns using a pairwise cross-loss and refined phase relation to maintain physical consistency, and Pix2Phase, which reconstructs phase maps corresponding to the top surface, bottom surface, and total thickness variation. Experimental validation with a Fizeau interferometer demonstrates performance comparable to conventional approaches using up to 37 interferograms, while reducing systematic error accumulation and execution time by more than 22 times. Physical-constraint evaluation further confirms stable and physically consistent reconstruction, with only random residual fluctuations and a root-mean-square error of 0.0017 μm. These results highlight the potential of the proposed framework for efficient interferometric metrology in dynamic applications such as in-line and in-process inspection.
- 21 April 2026
- Published by Elsevier in CIRP Annals
- Vol. 75 (1) , 679-683
Robust full-field absolute optical thickness profiling of glass substrates combining deep learning and virtual wavelength scanning Fourier analysis
- 21 April 2026
- Published by Elsevier in CIRP Annals
- Vol. 75 (1) , 679-683
Robust full-field absolute optical thickness profiling of glass substrates combining deep learning and virtual wavelength scanning Fourier analysis
- 15 April 2026
- Published by Springer Nature in The International Journal of Advanced Manufacturing Technology
- Vol. 144 (3-4) , 2961-2986
Volumetric error compensation using weighted least-squares fitting method based on sensitivity analysis in 5-axis machine tool
- 15 April 2026
- Published by Springer Nature in The International Journal of Advanced Manufacturing Technology
- Vol. 144 (3-4) , 2961-2986
Volumetric error compensation using weighted least-squares fitting method based on sensitivity analysis in 5-axis machine tool
- 7 April 2026
- Published by Springer Nature in International Journal of Precision Engineering and Manufacturing-Green Technology
- Vol. 13 (3) , 1431-1462
Recent Advances in Deep Learning-Based Interferometric Measurement
- 7 April 2026
- Published by Springer Nature in International Journal of Precision Engineering and Manufacturing-Green Technology
- Vol. 13 (3) , 1431-1462
Recent Advances in Deep Learning-Based Interferometric Measurement