• Smart Structures and Systems
• /
• 제33권1호
• /
• pp.55-64
• /
• 2024

This paper proposes a fusion imaging-based coating-defect classification method for steel structures that uses zero-shot learning. In the proposed method, a halogen lamp generates heat energy on the coating surface of a steel structure, and the resulting heat responses are measured by an infrared (IR) camera, while photos of the coating surface are captured by a charge-coupled device (CCD) camera. The measured heat responses and visual images are then analyzed using zero-shot learning to classify the coating defects, and the estimated coating defects are visualized throughout the inspection surface of the steel structure. In contrast to older approaches to coating-defect classification that relied on visual inspection and were limited to surface defects, and older artificial neural network (ANN)-based methods that required large amounts of data for training and validation, the proposed method accurately classifies both internal and external defects and can classify coating defects for unobserved classes that are not included in the training. Additionally, the proposed model easily learns about additional classifying conditions, making it simple to add classes for problems of interest and field application. Based on the results of validation via field testing, the defect-type classification performance is improved 22.7% of accuracy by fusing visual and thermal imaging compared to using only a visual dataset. Furthermore, the classification accuracy of the proposed method on a test dataset with only trained classes is validated to be 100%. With word-embedding vectors for the labels of untrained classes, the classification accuracy of the proposed method is 86.4%.

• 제어로봇시스템학회논문지
• /
• 제21권2호
• /
• pp.81-86
• /
• 2015

GPS position errors were corrected for guidance of autonomous vehicles. From the vision, we can obtain the lateral distance from the center of lane and the angle difference between the left and right detected line. By using a controller which makes these two measurements zero, a lane following system can be easily implemented. However, the problem is that if there's no lane, such as crossroad, the guidance system of autonomous vehicle does not work. In addition, Line detection has problems working on curved areas. In this case, the lateral distance measurement has an error because of a modeling mismatch. For this reason, we propose GPS error correction filter based on curve-modeled lane detection and evaluated the performance applying it to an autonomous vehicle at the test site.

• 제어로봇시스템학회논문지
• /
• 제4권3호
• /
• pp.384-391
• /
• 1998

When a device is mounted on the PCB, it is impossible to have zero defects due to many unpredictable problems. Among these problems, devices with bent corner leads due to mis-handling and which are not placed at a given point measured along the axis are principal problem in SMT(Surface Mounting Technology). It is obvious that given the complexity of the inspection task, the efficiency of a human inspection is questionable. Thus, new technologies for inspection of SMD(Surface Mounting Device) should be explored. An example of such technologies is the Automated Visual Inspection(AVI), wherein the vision system plays a key role to correct this problem. In implementing vision system, high-speed and high-precision are indispensable for practical purposes. In this paper, a new algorithm based on the Radon transform which uses a projection technique to inspect the FIC(Flat Integrated Circuit) device is proposed. The proposed algorithm is compared with other algorithms by measuring the position error(center and angle) and the processing time for the device image, characterized by line scan camera.

• Journal of Astronomy and Space Sciences
• /
• 제33권4호
• /
• pp.273-278
• /
• 2016

December 2009 was one of the quietest (monthly Ap=2) months over the last eight decades. It provided an excellent opportunity to study the day-to-day variability of the F2 layer with the smallest contribution due to geomagnetic activity. With this aim, we analyze hourly values of the F2-layer critical frequency (foF2) recorded at 18 ionosonde stations during the magnetically quietest (Ap=0) days of the month. The foF2 variability is quantified as the relative standard deviation of foF2 about the mean of all the "zero-Ap" days of December 2009. This case study may contribute to a more clear vision of the F2-layer variability caused by sources not linked to geomagnetic activity. In accord with previous studies, we find that there is considerable "zero-Ap" variability of foF2 all over the world. At most locations, foF2 variability is presumably affected by the passage of the solar terminator. The patterns of foF2 variability are different at different stations. Possible causes of the observed diurnal foF2 variability may be related to "meteorological" disturbances transmitted from the lower atmosphere or/and effects of the intrinsic turbulence of the ionosphere-atmosphere system.

• 로봇학회논문지
• /
• 제19권3호
• /
• pp.293-310
• /
• 2024

In this paper, we propose an agent model for Language-Driven Zero-Shot Object Navigation (L-ZSON) tasks, which takes in a freeform language description of an unseen target object and navigates to find out the target object in an inexperienced environment. In general, an L-ZSON agent should able to visually ground the target object by understanding the freeform language description of it and recognizing the corresponding visual object in camera images. Moreover, the L-ZSON agent should be also able to build a rich spatial context map over the unknown environment and decide efficient exploration actions based on the map until the target object is present in the field of view. To address these challenging issues, we proposes AML (Agent Model for L-ZSON), a novel L-ZSON agent model to make effective use of AI foundation models such as Large Language Model (LLM) and Vision-Language model (VLM). In order to tackle the visual grounding issue of the target object description, our agent model employs GLEE, a VLM pretrained for locating and identifying arbitrary objects in images and videos in the open world scenario. To meet the exploration policy issue, the proposed agent model leverages the commonsense knowledge of LLM to make sequential navigational decisions. By conducting various quantitative and qualitative experiments with RoboTHOR, the 3D simulation platform and PASTURE, the L-ZSON benchmark dataset, we show the superior performance of the proposed agent model.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1987년도 전기.전자공학 학술대회 논문집(II)
• /
• pp.1077-1080
• /
• 1987

Images are 2-dimensional projection of 3-dimensional scenes and many problems of scene analysis arise due to inherent depth ambiguities in a monocular 2-D image. Therefore, depth recovery is a crucial problem in image understanding. This paper proposes modified algorithm which is focused on accurate correspondnce in stereo vision. The feature we use is zero-crossing points and the similarity measure with two property evaluation function is used to estimate initial probability. And we introduce relaxation technique for accurate and global correspondence.

• Journal of Magnetics
• /
• 제19권3호
• /
• pp.215-220
• /
• 2014

This study investigates the dynamic magneto-mechanical behavior of magnetization-graded ferromagnetic materials Terfenol-D/FeCuNbSiB (MF). We measure the dynamic magneto-mechanical properties as a function of the DC bias magnetic field ($H_{dc}$). Our experimental results show that these dynamic magneto-mechanical properties are strongly dependent on the DC bias magnetic field. Furthermore, the dynamic strain coefficient, electromechanical resonance frequency, Young's moduli, and mechanical quality factor of Terfenol-D/FeCuNbSiB are greater than those of Terfenol-D under a lower DC bias magnetic field. The dynamic strain coefficient increases by a factor of between one and three, under the same DC bias magnetic field. In particular, the dynamic strain coefficient of Terfenol-D/FeCuNbSiB at zero bias achieves 48.6 nm/A, which is about 3.05 times larger than that of Terfenol-D. These good performances indicate that magnetization-graded ferromagnetic materials show promise for application in magnetic sensors.

• 동력기계공학회지
• /
• 제6권1호
• /
• pp.96-101
• /
• 2002

In this paper, we address the problem of controlling an end-effector to track and grab a moving target using the visual servoing technique. A visual servo mechanism based on the image-based servoing principle, is proposed by using visual feedback to control an end-effector without calibrated robot and camera models. Firstly, we consider the control problem as a nonlinear least squares optimization and update the joint angles through the Taylor Series Expansion. And to track a moving target in real time, the Jacobian estimation scheme(Dynamic Broyden's Method) is used to estimate the combined robot and image Jacobian. Using this algorithm, we can drive the objective function value to a neighborhood of zero. To show the effectiveness of the proposed algorithm, simulation results for a six degree of freedom robot are presented.

• 한국멀티미디어학회논문지
• /
• 제22권4호
• /
• pp.432-442
• /
• 2019

Most current human action recognition methods based on deep learning methods. It is required, however, a very high computational cost. In this paper, we propose an action change detection method to reduce repetitive human action recognition tasks. In reality, simple actions are often repeated and it is time consuming process to apply high cost action recognition methods on repeated actions. The proposed method decides whether action has changed. The action recognition is executed only when it has detected action change. The action change detection process is as follows. First, extract the number of non-zero pixel from motion history image and generate one-dimensional time-series data. Second, detecting action change by comparison of difference between current time trend and local extremum of time-series data and threshold. Experiments on the proposed method achieved 89% balanced accuracy on action change data and 61% reduced action recognition repetition.

• Journal of Platform Technology
• /
• 제11권6호
• /
• pp.34-46
• /
• 2023

자동차 타이어의 휠과 트레드 사이에 탄성중합체 또는 다각형의 스포크를 채우는 방식으로 제작하는 비공기압 타이어는 자동차 관련 학계 및 항공우주 업계의 중요한 연구 주제가 되고 있다. 본 연구에서는 생성형 적대 신경망을 기반으로 비공기압 타이어 디자인을 생성하는 시스템 개발했다. 특히 비공기압 타이어의 종류와 사용 환경, 제작 방식, 공기압 타이어와의 차이점 그리고 스포크 디자인에 따른 하중 전달의 변화 등 디자인에 영향을 미칠만한 변수들에 대한 조사를 실시했다. 이 연구는 OpenCV를 통해 다양한 스포크 형태의 이미지를 만들고, projected GANs에 학습시켜 비공기압 타이어 디자인에 사용될 스포크를 생성했다. 디자인된 비공기압 타이어는 사용 가능 및 불가능으로 레이블링하고, 이를 Vision Transformer 이미지 분류 AI 모델에 학습시켜 분류하도록 하였다. 최종적으로 분류 모델의 평가를 통해 0에 가까운 loss의 수렴, 99%의 정확도를 확인했다. 차후 도형 및 스포크 이미지와 알고리즘을 이용한 디자인이 아닌, 완전 자동화 시스템의 개발과 더 나아가 3D의 물리적 해석 없이 사용 가능한 디자인을 생성하는 것을 목표로 한다.