Daegu Gyeongbuk Institute of Science and Technology
DGIST Library Institutional RepositoryNot a member yet
12664 research outputs found
Sort by
메트릭 학습을 통해 향상된 단안 3D 객체 감지
No full textMonocular 3D object detection, Autonomous driving, Recognition, Regression, Metric LearningMonocular 3D object detection poses a significant challenge due to the lack of depth information in RGB images. Many existing methods strive to enhance the object depth estimation performance by allocating additional parameters for object depth estimation, utilizing extra modules or data. In contrast, we introduce a novel metric learning scheme that encourages the model to extract depth-discriminative features regardless of the visual attributes without increasing inference time and model size. Our method employs the distance- preserving function to organize the feature space manifold in relation to ground-truth object depth. The proposed (, , )-quasi-isometric loss leverages predetermined pairwise distance restriction as guidance for adjusting the distance among object descriptors without disrupting the non-linearity of the natural feature manifold. Moreover, we introduce an auxiliary head for object-wise depth estimation, which enhances depth quality while maintaining the inference time. The broad applicability of our method is demonstrated through experiments that show improvements in overall performance when integrated into various baselines. The results show that our method consistently improves the performance of various baselines by 25.27% and 4.54% on average across KITTI and Waymo, respectively.|단안 3D 객체 감지는 RGB 이미지에서의 깊이 정보 부족으로 인해 상당한 어려움을 가지고 있습니다. 많은 기존 방법들은 객체 깊이 추정 성능을 향상시키기 위해 추가 매개변수를 할당하거나 추가 모듈 또는 데이터를 활용하여 시도하고 있습니다. 이와 반대되게 본 논문에서는 시각적 특성을 고려하지 않고도 모델이 깊이가 잘 구별되는 특징을 추출하도록 하는 새로운 메트릭 학습 체계를 제안합니다. 이 방법은 추론 시간과 모델 크기를 늘리지 않으면서 모델이 시각적 특성과 상관없이 향상된 깊이 특징을 추출하도록 하는 거리 보존 함수를 활용하였다. 제안된 (, , ) -준등거리 손실함수는 객체 쌍 거리의 제한 조건을 활용하여 객체마다의 거리를 객체 깊이 라벨을 통해 정렬하며, 국소적으로 작용함으로써 고차원 공간에 존재하는 특징 매니폴드의 비선형성을 깨뜨리지 않습니다. 또한, 객체별 깊이 추정을 위한 보조 헤드를 도입함으로써 추론 시간을 증가시키지 않고 객체 깊이의 품질을 향상시켰습니다. KITTI 및 Waymo 데이터셋에서의 실험 결과는 다양한 기준선에 걸쳐 일관된 성능 향상을 보여주며 제안된 방법들의 효과를 강조합니다. 향후 작업으로, 우리의 방법은 다중 카메라 3D 객체 감지 시나리오 및 여러 하위 작업을 포함하는 다른 회귀 작업으로 확장될 가능성이 있습니다.I. Introduction 1
II. Related Work 2
III. Method 3
3.1 Preliminary 3
3.2 Problem Definition 4
3.3 Methodology 5
IV. Experiments 10
4.1 Evaluation Results on KITTI/Waymo Datasets 12
4.2 Additional Experiments 12
V. Theoretical Analysis 17
5.1 Pseudo-geodesic 17
5.2 Non-linearity Preservation via Local-constraint 19
VI. Conclusion 20
VII. References 21
VIII. 요약문 25MasterdCollectio
Adaptive transfer learning-based cryptanalysis on double random phase encoding
No full textEncrypting data can convert original data into a form that cannot be recognized and conceal personal information. To ensure data security, analyzing cryptographic algorithms used in data encryption is critical. The previous works evaluated the robustness of the optical cryptographic algorithm, Double Random Phase Encoding (DRPE), and verified the vulnerability of DRPE to plaintext recovery attacks based on deep learning. Although they could prove that DRPE is not secure by recovering the DRPE images into the original images, they evaluated DRPE only on simple datasets, which were binary or grayscale images with a single object. Thus, in this paper, we investigated the previous works of plaintext recovery on DRPE by using complex datasets. To our knowledge, it is the first study to evaluate DRPE on complex datasets. In addition to the plaintext recovery, we classified DRPE images of complex datasets and verified the feasible points in DRPE for classification tasks based on deep learning. Adaptive transfer learning, the modified version of the classical transfer learning for DRPE image classification, was proposed and utilized to train the DRPE image classification models. Experimental results showed that recovering the DRPE images of the complex datasets was challenging. In contrast, the imperceptible features representing the class of the images were in DRPE images regardless of whether the dataset was simple or complex. The classification performance of the proposed DRPE image classification scheme was better than those of state-of-the-art classification schemes, even with a small dataset. Furthermore, data augmentation with the proposed scheme improved the classification accuracy in DRPE image classification on a small dataset. In addition, our DRPE image classification schemes were applied to encrypted images by other optical cryptographic algorithms and the classification results were compared with that of the encrypted images by classical DRPE. © 2023 Elsevier LtdFALS
Recent Developments in Metallic Degradable Micromotors for Biomedical and Environmental Remediation Applications
No full textSynthetic micromotor has gained substantial attention in biomedicine and environmental remediation. Metal-based degradable micromotor composed of magnesium (Mg), zinc (Zn), and iron (Fe) have promise due to their nontoxic fuel-free propulsion, favorable biocompatibility, and safe excretion of degradation products Recent advances in degradable metallic micromotor have shown their fast movement in complex biological media, efficient cargo delivery and favorable biocompatibility. A noteworthy number of degradable metal-based micromotors employ bubble propulsion, utilizing water as fuel to generate hydrogen bubbles. This novel feature has projected degradable metallic micromotors for active in vivo drug delivery applications. In addition, understanding the degradation mechanism of these micromotors is also a key parameter for their design and performance. Its propulsion efficiency and life span govern the overall performance of a degradable metallic micromotor. Here we review the design and recent advancements of metallic degradable micromotors. Furthermore, we describe the controlled degradation, efficient in vivo drug delivery, and built-in acid neutralization capabilities of degradable micromotors with versatile biomedical applications. Moreover, we discuss micromotors’ efficacy in detecting and destroying environmental pollutants. Finally, we address the limitations and future research directions of degradable metallic micromotors.[Figure not available: see fulltext.] © 2023, The Author(s).TRU
Selective hydrocarbon or oxygenate production in CO2 electroreduction over metallurgical alloy catalysts
No full textAlloying of metals can be used to optimize intermediate binding during electrocatalysis but challenges remain in overcoming thermodynamic atomic miscibility in alloys. Here we report a coordination-controlled metal alloy in which copper clusters are spatially dispersed in a crystalline silver lattice to promote the electrochemical reduction of CO2 to ethanol. The synergistic interactions between Cu–Cu sites and Cu–Ag interfaces achieve highly selective hydrocarbon and oxygenate production by strengthening and diversifying the binding of *CO intermediates on terrace and defect sites. To control atomic coordinates beyond the miscibility limit and optimize the catalyst microstructure, sacrificial elements are incorporated with thermodynamically guided compositions to form intermetallic compounds. The sacrificial elements are then selectively dealloyed. Using a membrane electrode assembly, ethylene-selective production on copper catalysts (Faradaic efficiency, 69.6 ± 1.3%; full cell efficiency, 23.5%) is steered to ethanol-selective production on the supersaturated Ag–Cu solid-solution catalyst (Faradaic efficiency, 40.4 ± 2.4%; full cell efficiency, 14.4%). Metallurgy-designed catalyst fabrication enables the efficient chemical manufacturing of either hydrocarbons or oxygenates and offers guidelines for catalyst design principles. [Figure not available: see fulltext.]. © 2023, The Author(s), under exclusive licence to Springer Nature Limited.FALSEscopu
Robotic Skin Mimicking Human Skin Layer and Pacinian Corpuscle for Social Interaction
No full textHumans sense and interpret touches on their skin for social interaction. Similarly, robotic systems with human-like robotic skin can intuitively interact with humans. Therefore, many tactile sensors have been developed, but their excessive sensing elements, narrow sensing bandwidth, and fragility limit their applications as robotic skin. This article proposes a robotic skin structure that mimics the human skin layer and Pacinian corpuscle using a textured resilient fabric, a uniquely structured airmesh, and encapsulated microphones. Furthermore, functions such as tactile stimulus encoding, tactile stimulus dispersion, elastomechanical properties, and wide sensitivity bandwidth are mimicked. The developed skin identifies tactile locations and patterns using a small number of sensing nodes and algorithms that interpret tactile sensations. These include passive acoustic tomography to localize touch, signal intensities map, and spectrogram to encode spatiotemporal characteristics of touch, and convolutional neural network to decode and classify touch. As a result, the algorithms localized tactile stimulus with a mean error of 1.8 cm and classified touch into nine classes with an accuracy of 93.3%. Furthermore, the developed robotic skin has no rigid material and employs a few sensing nodes, thus easily accommodating large nonplanar surfaces. The skin was implemented on a robotic arm to demonstrate a physical human–robot interaction and on a vertically cylindrical surface of similar size to a social robot to demonstrate the scaling up to larger systems with a lower sensing node density. © 2023 IEEEFALSEsciescopu
Assembled elastic actuator with interchangeable modules
No full text본 발명은 외부에서 인가되는 전원으로 회전축을 회전시켜 회전력을 발생하는 모터모듈; 상기 모터모듈의 상부에 적층되면서 상기 모터모듈의 회전축과 입력측이 연결되어, 상기 모터모듈에서 인가된 토크를 증폭하여 출력측으로 증폭된 토크를 출력하고, 상기 모터모듈에서 인가된 토크와 외부에서 작용하는 토크로 인해 기어하우징이 회전할 수 있는 기어모듈; 하측에 상기 모터모듈을 내장하면서 상기 모터모듈과 결합하고, 대향진 상측에 상기 기어모듈을 내장 가능하면서 상기 기어모듈과 결합하며, 반발 토크 및 외부에서 인가되는 외력에 의해 변형되는 탄성부재; 상기 모터모듈 및 기어모듈이 내장된 탄성부재를 내부에 수용하면서 하측은 상기 모터모듈의 주변과 결합되고, 대향진 상측은 상기 기어모듈이 회전가능하게 연결되는 로드케이스; 및 중앙에 관통홀이 형성된 환상으로, 상기 관통홀을 통해 상기 기어모듈의 출력측을 외부로 노출시키면서 상기 로드케이스의 상측단에 결합되는 커버;를 포함하고, 상기 커버의 면적을 따라 복수 개의 강성 전환용 잠금홀을 형성하고, 상기 강성 전환용 잠금홀을 통해 볼트가 기어모듈의 기어하우징에 체결되면 강성 구동기로 전환되도록 하여, 탄성 구동기를 이루는 개별 구성요소(모터모듈, 기어모듈, 탄성부재, 로드케이스, 커버)가 각각 독립된 개별 모듈을 조립하는 형태로 이루어져 있어, 특정 모듈을 모니터링할 시, 다른 모듈을 건드리지 않아도 되므로, 구동기의 유지 보수가 용이하고, 개별 모듈과 연결부위에 배치되는 크로스 롤러 베어링의 구조로 인해 구동기가 구조적으로 동심이 유지 및 보장되어 조립이 용이하며, 축 회전 방향을 제외한 방향으로부터 인가되는 힘과, 토크로부터 탄성부재가 영향을 받지 않아, 구동기의 효율 극대화 및 외력 추정의 정밀도와 정확성이 확보되며, 부품 교체 혹은 탄성 구동기의 해제과정 없이 탄성 구동기에서 일반적인 고강성 구동기로 전환 가능하여, 구동기의 용도 확장성이 향상되는 모듈 교체가 가능한 조립형 탄성 구동기를 제공한다
SPIN-ORBIT TORQUE MAGNETIC DEVICE CONTROLLED ON-OFF BASED ON ELECTRIC FIELD EFFECT
No full text본 발명은 스핀-궤도 토크 자기 소자 및 그 제조방법에 관한 것으로, 일실시예에 따른 스핀-궤도 토크 자기 소자는 제1 중금속층과, 제1 중금속층 상에 형성된 강자성층과, 강자성층 상에 형성된 제2 중금속층 및 제2 중금속층 상에 형성된 게이트 산화물층을 포함하고, 여기서, 제2 중금속층은 게이트 산화물층에 기설정된 크기의 게이트 전압이 인가되면, 스핀 궤도 상호작용의 강도가 제어될 수 있다
Li-ion hopping conduction enabled by associative Li-salt in acetonitrile solutions
No full textTo date, ionic conduction in nonaqueous electrolytes has been explained through the vehicle-type migration mechanism. Yet, new research hints at another conduction mode: ion-hopping, seen in highly concentrated solutions with multi-coordinating solvents. Our research uncovers that Li-ion hopping conduction also occurs in monodentate acetonitrile (AN) electrolytes, enabled by a highly associative Li-salt. Using techniques like pulse-field gradient NMR, Raman spectroscopy, and dielectric relaxation spectroscopy, we examined AN solutions with lithium trifluoroacetate (LiTFA) and lithium bis(fluorosulfonyl)imide (LiFSI). Results showed that Li-ion diffusion in LiTFA-AN was faster due to an anion-bridge structure formed by the associative nature of LiTFA. In contrast, the LiFSI-AN solution demonstrated slower Li-ion movement. In practical applications, like LiFePO4 symmetric cells, 4 M LiTFA-AN outperformed 1 M LiTFA-AN in rate performance, despite its lower ionic conductivity. This challenges the belief that associative Li-salts are unsuitable for battery electrolytes and prompts reconsideration of other associative Li-salts. © 2023 Korean Chemical Society, Seoul & Wiley-VCH GmbH.FALSEsciescopuskc
Revealing Two Distinct Formation Pathways of 2D Wurtzite-CdSe Nanocrystals Using In Situ X-Ray Scattering
Get PDFUnderstanding the mechanism underlying the formation of quantum-sized semiconductor nanocrystals is crucial for controlling their synthesis for a wide array of applications. However, most studies of 2D CdSe nanocrystals have relied predominantly on ex situ analyses, obscuring key intermediate stages and raising fundamental questions regarding their lateral shapes. Herein, the formation pathways of two distinct quantum-sized 2D wurtzite-CdSe nanocrystals — nanoribbons and nanosheets — by employing a comprehensive approach, combining in situ small-angle X-ray scattering techniques with various ex situ characterization methods is studied. Although both nanostructures share the same thickness of ≈1.4 nm, they display contrasting lateral dimensions. The findings reveal the pivotal role of Se precursor reactivity in determining two distinct synthesis pathways. Specifically, highly reactive precursors promote the formation of the nanocluster-lamellar assemblies, leading to the synthesis of 2D nanoribbons with elongated shapes. In contrast, mild precursors produce nanosheets from a tiny seed of 2D nuclei, and the lateral growth is regulated by chloride ions, rather than relying on nanocluster-lamellar assemblies or Cd(halide)2–alkylamine templates, resulting in 2D nanocrystals with relatively shorter lengths. These findings significantly advance the understanding of the growth mechanism governing quantum-sized 2D semiconductor nanocrystals and offer valuable guidelines for their rational synthesis. © 2023 The Authors. Advanced Science published by Wiley-VCH GmbH.TRUEsciescopu