Wonsuk Jung

To achieve precise control over smart material functionality, the effects of compositional variations on the network structure and thermo‐responsive behavior of PNIPAm‐PEGDA hydrogels were systematically investigated. This study establishes a quantitative correlation between monomer ratios and macro‐scale properties. As the PEGDA content increased, the enhanced cross‐linking density led to a more compact polymer network, resulting in a significant mechanical enhancement of up to 20.4 MPa, compared to 0.39 MPa for p1n9 composition (50‐fold increase). NIPAm‐rich compositions promoted chain mobility and hydration, exhibiting a temperature‐dependent length swelling increment ranging from 0.05 (high temperature) to 0.84 (low temperature), corresponding to a 17‐fold variation between contracted and swollen states. These structural alterations governed drug loading and kinetics under identical loading conditions. NIPAm‐rich hydrogels exhibited a substantially higher drug release capacity when normalized by dry mass (mg/mg), showing approximately a 7.3‐fold increase compared to PEGDA‐rich compositions. Furthermore, compositional tuning enabled the characteristic transition temperature to be shifted from 33°C to 45°C, providing a broad, programmable thermo‐responsive window. Observed trends were validated through Flory–Huggins interaction parameters and SEM analysis. Developed regression models provide a predictive framework for tailoring mechanical stiffness and release profiles, offering a systematic design guideline for functional hydrogels in drug delivery.

In this paper, we present a robust whole-body controller which enables wheeled-bipedal robots to achieve robust locomotion on various environments in the absence of terrain perception. The proposed controller combines a whole-body motion planner and an optimization-based torque generator. By accounting for the floating base dynamics of the robot, the whole-body motion planner generates the terrain-adaptive motion based on postural stabilizing forces and the zero moment point (ZMP) that maintains the robot’s balance without relying on terrain information. In addition, the optimization-based torque generator integrates a linear quadratic regulator (LQR) and a quadratic programming (QP)-based controller. The LQR controller generates the wheel torques using the wheeled inverted pendulum model (WIPM) to regulate the robot body’s angle while considering the inherent non-minimum phase behavior. Given the computed wheel torques, the hierarchical QP controller distributes the optimal joint torques corresponding to the desired motion to realize the stable balance of the robot. The proposed controller is tested on a wheeled-bipedal robot in the Gazebo simulator. Locomotion abilities on various terrains including slopes, stairs, and uneven terrains and under external forces validated in simulations.

Many studies have explored non-invasive glucose monitoring by measuring glucose levels in the aqueous humor of the anterior chamber, which correlate with blood glucose. However, most approaches rely on complex optical systems, and their robustness to inter-individual variability remains insufficiently examined. In this study, we demonstrate the feasibility of non-invasive aqueous humor glucose measurement using a hardware-simplified NIR spectroscopy setup integrated with machine learning. A photodiode-based system operating at four discrete wavelengths (1150, 1300, 1400, and 1620 nm) was evaluated using an artificial anterior chamber, with cornea and iris phantoms fabricated from PDMS and TiO₂. Glucose solutions ranging from 75 to 150 mg dl ⁻¹ were prepared. Machine learning models incorporating wavelength-ratio-based features were used to mitigate inter-iris variability. Under within-iris calibration, a Gradient Boosting model achieved an R ² of 0.983 and an RMSE of 3.45 mg dl ⁻¹ . Across multiple irises, the framework maintained an average prediction error of approximately 15 mg dl ⁻¹ , indicating robustness to inter-iris scattering variability under simplified hardware and calibration assumptions. These results indicate that reliable glucose estimation can be achieved without reliance on complex optical architectures within the evaluated experimental conditions, and suggest the potential of ratio-based learning frameworks as a foundation for non-invasive glucose sensing in ocular measurement settings.

Conventional photolithography-based patterning of monolayer graphene inevitably causes delamination and contamination when fabricating sub-5 μm patterns, rendering electrical characterization impossible. Here, we develop a one-step free patterning of graphene (OFP-G) method, which selectively converts C = C/C-C bonds into C-O bonds, enabling robust micro-sized pattern fabrication without delamination. OFP-G achieves intact 5 μm and 20 μm-wide patterns with resistances of 11.5 ± 2.8 Ω and 9.4 ± 0.4 Ω, respectively, overcoming the inherent limitations of photolithographic approaches that induce graphene delamination and contamination. Molecular dynamics simulations reveal that C-O bond conversion significantly reduces interfacial strain, while Raman and XPS analyses confirm that the patterned regions maintain structural integrity without etching-induced defects. By eliminating photoresist contamination and damage, this approach provides a scalable and reliable alternative for high-resolution graphene micro-sized pattern fabrication.

A key challenge in material delivery systems is ensuring the enhanced durability to prevent moisture loss and controlled dispersion of capsules carrying agents. This study introduces octadecylamine-functionalized graphene oxide with carbonyl iron particles (ODA-GO/CIP) capsules, synthesized via a microfluidic technique to enhance moisture retention and magnetic responsiveness. X-ray photoelectron spectroscopy and scanning electron microscopy confirmed successful ODA-GO/CIP integration. Experiments showed an 8% increase in magnetic properties and a 693-fold improvement in water storage compared to pure CIP capsules. Under a magnetic field, these capsules exhibited significantly enhanced dispersion.

We report the fabrication of lightweight porous carbon structures via UV-assisted photopolymerization molding using a commercial photocurable resin modified with natural tannin and sodium carboxymethyl cellulose (NaCMC) as sustainable additives. A systematic analysis was conducted by applying a Design of Experiments (DOE) approach and regression modeling to evaluate the effects of varying blend compositions on carbon yield and mechanical strength. The results indicate that increasing the tannin content led to a maximum carbon yield of 13.43%, with an average porosity of approximately 80% and a compressive strength around 1 kPa. NaCMC was found to effectively control the resin viscosity within printable limits of 0.2537 Pa·s, although NaCMC indirectly improved carbonization efficiency through normalized yield analysis. This work highlights the synergistic role of bio-based polymers in tuning porous carbon properties. The findings provide a data-driven framework for designing sustainable polymer-derived carbon materials, bridging additive manufacturing with green chemistry.

기업벤처투자는 재무적 이익 실현을 주목적으로 하는 전문벤처투자와 달리 모기업의 전략적 목적을 달성하기 위한 방안으로 발전되고 있다. 다양한 기업 환경에 따른 다변화된 투자 형태가 진행되고 있으므로 이에 대한 연구가 필요하다. 본 논문은 모기업과 투자대상 벤처기업의 환경이 기업벤처투자 형태 중 인수 및 합병에 미치는 영향을 연구한다. 기존 연구 문헌을 통해서 기업벤처투자에 영향을 미치는 주요 기업 환경요인으로 모기업의 산업군, 모기업과 투자대상 벤처기업과의 사업영역 거리를 도출하고 기업벤처투자 유형인 인수(acquisitions)와 합병(mergers)과의 관계를 로지스틱 회귀분석으로 분석하였다. 분석 결과, 모기업과 투자 대상 벤처기업의 산업군이 일치할 때와 모기업과 투자 대상 벤처기업의 지리적 거리가 가까울수록 인수보다 합병을 선호했다. 본 연구 결과는 기업벤처투자에 대한 중요성 및 확대가 이루어지고 있는 상황에서 최근 데이터를 활용하여 연구모형을 검증했다는 점에서 의의가 있으며, 실무적으로 기업벤처투자의 전략 및 운영방식에 대한 가이드라인을 제공할 것이다.

The use of global magnetic fields—typically produced by pairs of Maxwell and Helmholtz coils—offers the advantage of generating uniform magnetic fields and magnitude uniform gradient distributions over extensive volumes, while ensuring easy control. However, achieving selective control of multiple magnetic microrobots under such global fields remains problematic. In this study, we utilize a magnetic focus field generated by a pair of Maxwell coils to impede the magnetic torque-based movement of magnetic robots confined within channels. Although the magnetic field is null at the focus field’s center, it increases linearly in all directions. This exerts magnetic forces that push robots, which are situated away from the center, toward the channel walls. This not only restricts their positioning but also reduces their responsiveness to additional fields. By introducing uniform dc magnetic fields to the focus field, we can change the controlled robot. This mechanism for selective locomotion has been empirically validated with the selective movement of two helical magnetic robots and swarms of magnetic microparticles. Furthermore, leveraging the same focus field, we present a novel separation mechanism for magnetic particle swarms, enhancing the aforementioned selective locomotion mechanism. The practical implications of our proposed locomotion mechanism have been showcased in targeted delivery, drilling, and improved magnetic heating experiments.

In recent years, self-healing research has been actively carried out to increase the life and stability of concrete structures. A representative method of concrete self-healing is a method using a bacterial substrate and a capsule. Bacteria-based self-healing is greatly influenced by the external environment, and sleeping structures are more suitable for capsule-based methods. In this study, the mechanical properties of the bonding surface of the photocurable resin are used to grow cracks in the capsule so that the self-healing capsule pops out and the healing liquid inside the capsule can be diffused efficiently. We proceeded with research to form pathways. The study analyzed the concordance rate of the bonding surface of the capsule and the judging position, and analyzed the factors affecting the concordance rate. As a result, by adjusting the number of bonding surfaces and the degree of air bubble removal, the rate of coincidence between the bonding surfaces and the fracture position was increased from 15.48% to 63.18% by a factor of 4.08.

This paper is concerned with the syntax and semantics of [muy amigo-possessive] in Spanish. Like predicative nouns and adjectives in the complement position of the copulative verb ser, the string of [muy amigo-possessive] indicates a property of the subject, and at the same time shows a true possessive relationship between the word amigo and the postnominal possessive. To capture these properties structurally, I present an analysis of [muy amigo-possessive] that accounts for the categorial relationship between the word amigo and the postnominal possessive. Specifically, I argue that the word amigo contained in the string of [muy amigo-possessive] undergoes a recategorization from noun to adjective in order to be consistent with categorial selection. From the perspective of Distributed Morphology, I propose that the word amigo, which is introduced as a category-neutral root amig, merges first with a nominalizer to yield the structure of [amigo-possessive], and then merges with a adjectivizer as a process of recategorization to finally produce the structure of [muy [amigo-possessive]]. Under this analysis, categorial selection of the constituents in the string of [muy amigo-possessive] is satisfied in an appropriate way, and the syntactic distribution of each constituent is also explained. Finally, a number of alternatives to the proposed analysis are critically examined.