• Proceedings of the Korean Institute of Surface Engineering Conference
• /
• 2018.06a
• /
• pp.14-14
• /
• 2018

• The Journal of Korea Robotics Society
• /
• v.15 no.2
• /
• pp.197-204
• /
• 2020

In this paper, we present a learning platform for robotic grasping in real world, in which actor-critic deep reinforcement learning is employed to directly learn the grasping skill from raw image pixels and rarely observed rewards. This is a challenging task because existing algorithms based on deep reinforcement learning require an extensive number of training data or massive computational cost so that they cannot be affordable in real world settings. To address this problems, the proposed learning platform basically consists of two training phases; a learning phase in simulator and subsequent learning in real world. Here, main processing blocks in the platform are extraction of latent vector based on state representation learning and disentanglement of a raw image, generation of adapted synthetic image using generative adversarial networks, and object detection and arm segmentation for the disentanglement. We demonstrate the effectiveness of this approach in a real environment.

• Proceedings of the KSME Conference
• /
• 2008.11b
• /
• pp.1984-1989
• /
• 2008

In this study, fire simulations were performed to analyze the characteristics of the fire driven flow and the effects of the platform screen door on the smoke flow in the station, when the fire occurred in the center of the platform. Soongsil Univ. station (line number 7, 47m in depth underground) was chosen which was the one of the deepest underground subway stations in the Seoul metro, SMRT. The parallel computational method was employed to compute the heat and mass transfer eqn's with 6 CPUs of the linux clustering machine. The fire driven flow was simulated with using FDS code in which LES method was applied. The Heat release rate was 10MW and The Ultrafast model was applied for the growing model of the fire source. The 10,000,000 structured grids were used.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.21 no.11
• /
• pp.884-891
• /
• 2020

The collapse of bridges can cause massive casualties and economic losses. Therefore, it is thus essential to evaluate the structural safety of bridges. For this task, structural reliability analysis, considering various bridge-related uncertainty factors, is often used. This paper proposes a new computational platform to perform structural reliability analysis for bridges and evaluate their structural safety under various loading conditions. For this purpose, a software package of reliability analysis, Finite Element Reliability Using MATLAB (FERUM), was integrated with MIDAS/CIVIL, which is a widely-used commercial software package specialized for bridges. Furthermore, a graphical user interface (GUI) control module has been added to FERUM to overcome the limitations of software operation. In this study, the proposed platform was applied to a simple frame structure, and the analysis results of the FORM (First-Order Reliability Method) and MCS (Monte Carlo simulation), which are representative reliability analysis methods, were compared. The proposed platform was verified by confirming that the calculated failure probability difference was less than 5%. In addition, the structural safety of a pre-stressed concrete (PSC) bridge was evaluated considering the KL-510 vehicle model. The proposed new structural reliability analysis platform is expected to enable an effective reliability-based safety evaluation of bridges.

• Journal of Computing Science and Engineering
• /
• v.7 no.3
• /
• pp.198-203
• /
• 2013

In this paper, we focus on a novel Internet architecture, based on the urbanization of virtual machines. In this approach, virtual networks are built linking specific virtual elements (router, switch, firewall, box, access point, etc.). A virtual network represents a network with an independent protocol stack that shares resources from the underlying network infrastructure. Virtualization divides a real computational environment into virtual computational environments that are isolated from each other, and interact with the upper computational layer, as would be expected from a real, non-virtualized environment. Metamorphic networks enhance several concepts related to future networks, and mainly the urbanization of virtual machines. We present this new paradigm, and the methodology, based on the worldwide metamorphic network platform "M-Net". The metamorphic approach could solve many complex problems, especially related to Cloud computing services.

• Journal of computational fluids engineering
• /
• v.15 no.3
• /
• pp.46-53
• /
• 2010

Recently, several open source codes for computational fluid dynamics (CFD) have been introduced and are spreading fast. Our group has chosen the OpenFOAM as a platform to develop our own in-house code. In this brief review, we would like to share the information on the codes and what we have experienced so far. We introduce several features of OpenFOAM, which include the performance compared with commercial packages, estimation for current user population, and our own prospect for future improvement in performance and growth in user population. In addition, we briefly introduce our experience gained in embedding the level set method into the OpenFOAM.

• Journal of the Korean institute of surface engineering
• /
• v.50 no.1
• /
• pp.1-9
• /
• 2017

Design of novel materials in renewable energy systems plays a key role in powering transportation vehicles and portable electronics. This review introduces the research work of first principles-based computational design for the materials over the last decade to accomplish the goal with less financial and temporal cost beyond the conventional approach, especially, focusing on electrocatalyst toward a proton exchange membrane fuel cell (PEMFC). It is proposed that the new method combined with experimental validation, can provide fundamental descriptors and mechanical understanding for optimal efficiency control of a whole system. Advancing these methods can even realize a computational platform of the materials genome, which can substantially reduce the time period from discovery to commercialization into markets of new materials.

• Proceedings of the Computational Structural Engineering Institute Conference
• /
• 2011.04a
• /
• pp.639-642
• /
• 2011

이 연구에서는 완전 조립식 교량 하부구조의 비선형해석을 위한 전산플랫폼을 개발하였다. 완전 조립식 교량 하부구조의 비선형거동을 정확하게 파악하고 합리적이면서 경제적인 설계기준의 개발을 위한 자료를 제공하는데 그 목적이 있다. 재료적 비선형성에 대해서는 균열콘크리트에 대한 인장, 압축, 전단모델과 콘크리트 속에 있는 철근모델을 조합하여 고려하였다. 사용된 부착 또는 비부착 텐던요소는 유한요소법에 근거하며 프리스트레스트 콘크리트 부재의 콘크리트와 텐던의 상호작용을 구현할 수 있다. 그리고 접합면요소는 세그먼트 접합부의 비탄성거동을 예측할 수 있다. 제안된 해석기법은 수치예제에 대하여 비선형거동을 비교적 정확하게 예측하였다.

• Proceedings of the Computational Structural Engineering Institute Conference
• /
• 2011.04a
• /
• pp.504-507
• /
• 2011

3D graphic softwares have brought design spaces beyond the limitations of Euclidean space. Moreover, as computational geometry has been considered together with algorithms, generative algorithms are being evolved. Recently 3D graphic softwares with the embedded generative algorithms allow designers to design free form curves and surfaces in a systematic way. While architectural design has been greatly affected by the advancement of 3D graphic technology, such attention has not given in the realm of structural design. Grasshopper is a platform in Rhino to deal with these Generative Algorithms and Associative modelling techniques. This study has tried to develop a module for preliminary structural configuration using Rhino with Grasshopper. To verify the proposed concept in this study, a module for designing a basic type of suspension structure is introduced.

• Proceedings of the Computational Structural Engineering Institute Conference
• /
• 2010.04a
• /
• pp.734-737
• /
• 2010

이 연구에서는 프리스트레스트 콘크리트 깊은 보의 비선형해석을 위한 전산플랫폼을 개발하였다. 프리스트레스트 콘크리트 깊은 보의 전단거동을 정확하게 파악하고 합리적이면서 경제적인 설계기준의 개발을 위한 자료를 제공하는데 그 목적이 있다. 재료적 비선형성에 대해서는 균열콘크리트에 대한 인장, 압축, 전단모델과 콘크리트 속에 있는 철근모델을 조합하여 고려하였다. 사용된 부착 또는 비부착 텐던요소는 유한요소법에 근거하며 프리스트레스트 콘크리트 부재의 콘크리트와 텐던의 상호작용을 구현할 수 있다. 이 연구에서는 프리스트레스트 콘크리트 깊은 보의 전단거동을 파악하기 위해 제안한 해석기법을 신뢰성 있는 연구자의 실험 결과와 비교하여 그 타당성을 검증하였다.