• 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.

• 정보과학회 논문지
• /
• 제43권12호
• /
• pp.1365-1375
• /
• 2016

일상생활 환경 속에서 자율적으로 동작하는 서비스 로봇에게 가장 필수적인 능력 중 하나가 동적으로 변화하는 주변 환경에 대한 올바른 상황 인식과 이해 능력이다. 다양한 센서 데이터 스트림들로 부터 신속히 의사 결정에 필요한 고수준의 상황 지식을 생성해내기 위해서는, 멀티 모달 센서 데이터의 융합, 불확실성 처리, 기호 지식의 실체화, 시간 의존성과 가변성 처리, 실시간성을 만족할 수 있는 시-공간 추론 등 많은 문제들이 해결되어야 한다. 이와 같은 문제들을 고려하여, 본 논문에서는 지능형 서비스 로봇을 위한 효과적인 동적 상황 관리 및 시-공간 추론 방법을 제시한다. 본 논문에서는 상황 지식 관리와 추론의 효율성을 극대화하기 위해, 저수준의 상황 지식은 센서 및 인식 데이터가 입력될 때마다 실시간적으로 생성되지만, 반면에 고수준의 상황 지식은 의사 결정 모듈에서 요구가 있을 때만 후향 시-공간 추론을 통해 유도되도록 알고리즘을 설계하였다. Kinect 시각 센서 기반의 Turtlebot를 이용한 실험을 통해, 제안한 방법에 기초한 동적 상황 관리 및 추론 시스템의 높은 효율성을 확인할 수 있었다.

• 한국산업정보학회논문지
• /
• 제6권3호
• /
• pp.51-64
• /
• 2001

스키마 통합은 이종 분산(Heterogeneous and Distributed) 지리데이타베이스 시스템 (GDS, Geographic Database Systems)에 있어서 해결해야 할 가장 과제들 중의 하나이다. 다양한 응용분야에 있어서 공간정보(spatial information)의 사용이 점차적으로 증가해 감에 따라 지리정보의 통합은 의사결정자들에게 있어 대단히 중요한 문제가 되었다. 그러나, 데이타베이스 관련 문헌에 기술되고 있는 기존의 스키마통합 기법은 시각적인 데이터, 공간정보, 임시적인 정보들을 내포하고 있는 복잡한 객체들간의 이질성(heterogeneity)의 관리라는 문제를 간과하고 있다. 스키마통합의 어려움은 의미(semantics)상의 혼돈뿐만 아니라 공간모형에 대한 상이한 표현으로부터도 초래된다. 그러므로, 지리데이타베이스 분야에서 데이터베이스간의 상호작용성(interoperability)을 실현하는 것은 생각했던 것보다 훨씬 복잡한 문제를 야기하게 되는 것이다. 본 연구에서는 이러한 문제의 해결을 시도하기 위하여 이종 분산 지리데이타베이스에 있어서 스키마통합을 지원할 수 있는 방법론과 프로토타입 도구를 소개하고자 한다.

• 한국통신학회논문지
• /
• 제38A권1호
• /
• pp.1-9
• /
• 2013

본 논문에서는 CS-LBP (Center-Symmetric Local Binary Pattern) 특징과 공간 피라미드를 이용한 BoF (Bag of Features)를 생성하고 이를 랜덤 포레스트(Random Forest) 분류기에 적용하여 인간의 행동을 인식하는 알고리즘을 제안한다. BoF를 생성하기 위해 영상을 균일한 패치로 나누고, 각 패치 마다 CS-LBP 특징을 추출한다. 행동 분류 성능을 향상시키기 위해 패치들마다 추출한 특징벡터들에 대해 K-mean 클러스터링을 적용하여 코드 북을 생성한다. 본 논문에서는 영상의 지역적인 특성을 고려하기 위해 공간 피라미드 방법을 적용하고 각 공간 레벨에서 추출된 BoF에 대해 가중치를 적용하여 최종적으로 하나의 특징 벡터로 결합한다. 행동 분류를 위해 결정트리의 앙상블로 이루어진 랜덤 포레스트는 학습 단계에서 각 행동 클래스를 위한 분류 모델을 만든다. 가중 BoF가 적용된 랜덤 포레스트는 다양한 인간 행동 영상을 포함하고 있는 Standford Actions 40 데이터를 성공적으로 분류하였다. 또한 기존 방법에 비해 분류 성능이 유사하거나 우수하며, 한 장의 영상에 대해 빠른 인식속도를 보였다.

• 한국경영과학회:학술대회논문집
• /
• 대한산업공학회/한국경영과학회 2006년도 춘계공동학술대회 논문집
• /
• pp.373-378
• /
• 2006

Recently the technical advances and complexities have generated much of the difficulties in managing the project resources, for both scheduling and costing to accomplish the project in the most efficient manner. The project manager is frequently required to render judgments concerning the schedule and resource adjustments. This research develops an analytical model for a schedule-cost and risk analysis based on visual PERT/CPM. We used a three-step approach: 1) in the first step, a deterministic PERT/CPM model for the critical path and estimating the project time schedule and related resource planning and we developed a heuristic model for crash and stretch out analysis based upon a time-cost trade-off associated with the crash and stretch out of the project. 2) In second step, we developed web-based risk evaluation model for project analysis. Major technologies used for this step are AHP (analytic hierarchy process, fuzzy-AHP, multi-attribute analysis, stochastic network simulation, and web based decision support system. Also we have developed computer programs and have shown the results of sample runs for an R&D project risk analysis. 3) We developed an optimization model for project resource allocation. We used AHP weighted values and optimization methods. Computer implementation for this model is provided based on GUI-Type objective-oriented programming for the users and provided displays of all the inputs and outputs in the form of GUI-Type. The results of this research will provide the project managers with efficient management tools.

• 한국실내디자인학회논문집
• /
• 제25권4호
• /
• pp.80-89
• /
• 2016

The analysis of spatial search direction and its concentration through eye movement can produce some useful data in that it enables to know the features of space elements and their effects on one another. The results by analysing the search features and concentration of spatial sections through the eye-tracking in shops in a department store makes it possible to define the followings. First, the features of 'eye's in & out' could be estimated through the division of sections by the characteristics of those shops and the extraction of central point based on the decision of continuative observation. The decision of continuative observations enabled to analyse the frequency of observation data which can be considered to be 'things watched longtime' and the stared points that is equivalent to 'things seen very often', by which the searching characteristics of spatial sections could be estimated. Second, as with the eye's [in], the right shops had 0.6 times more (3.5%) than those left and as with the eye's [out] the left ones had 0.6 times more (3.5%). It indicates that [in, out] of the right and the left shops had the same difference, which lets us know that with starting point of the middle space, [in] and [out] were paid more attention to the right shops and the left shops respectively. Third, as with the searching directions by section, the searching times [2.9 times] from [B] to [A] were than that [2.6 times] from [A] to [B]. It was also found that the left shops had more searching direction toward [C, D] than the right ones and that those searching activities at the left shops were more active. Fourth, when the searching directions by section are reviewed, the frequency of searching from [B] to [A] was 2.9 and that of the other way 2.6. Also the left shops were found to have more searching direction toward [C, D] than the right ones and those searching activities at the left shops were estimated to be more active.

• 경영과정보연구
• /
• 제4권
• /
• pp.257-284
• /
• 2000

According as the drivers increase who have their cars, the comprehensive studies on the automobile for the traffic safety have been raised as the important problems. Visual Recognition System for radio-controled driving is a part of the sensor processor of Unmanned Autonomous Vehicle System. When a driver drives his car on an unknown highway or general road, it produces a model from the successively inputted road traffic information. The suggested Recognition System of the Road Traffic Safety Information Board is to recognize and distinguish automatically a Road Traffic Safety Information Board as one of road traffic information. The whole processes of Recognition System of the Road Traffic Safety Information Board suggested in this study are as follows. We took the photographs of Road Traffic Safety Information Board with a digital camera in order to get an image and normalize bitmap image file with a size of $200{\times}200$ byte with Photo Shop 5.0. The existing True Color is made up the color data of sixteen million kinds. We changed it with 256 Color, because it has large capacity, and spend much time on calculating. We have practiced works of 30 times with erosion and dilation algorithm to remove unnecessary images. We drawing out original image with the Region Splitting Technique as a kind of segmentation. We made three kinds of grouping(Attention Information Board, Prohibit Information Board, and Introduction Information Board) by RYB( Red, Yellow, Blue) color segmentation. We minimized the image size of board, direction, and the influence of rounding. We also minimized the Influence according to position. and the brightness of light and darkness with Eigen Vector and Eigen Value. The data sampling this feature value appeared after building the learning Code Book Database. The suggested Recognition System of the Road Traffic Safety Information Board firstly distinguished three kinds of groups in the database of learning Code Book, and suggested in order to recognize after comparing and judging the board want to recognize within the same group with Nearest Neighborhood Decision Making.

• 대한치매학회지
• /
• 제22권4호
• /
• pp.148-157
• /
• 2023

Background and Purpose: Episodic memory is a system that receives and stores information about temporally dated episodes and their interrelations. Our study aimed to investigate the relevance of episodic memory to time perception, with a specific focus on simultaneity/order judgment. Methods: Experiment 1 employed the simultaneity judgment task to discern differences in time perception between patients with mild cognitive impairment or dementia, and age-matched normals. A mathematical analysis capable of estimating subjects' time processing was utilized to identify the sensory and decisional components of temporal order and simultaneity judgment. Experiment 2 examined how differences in temporal perception relate to performance in temporal order memory, in which time delays play a critical role. Results: The temporal decision windows for both temporal order and simultaneity judgments exhibited marginal differences between patients with episodic memory impairment, and their healthy counterparts (p = 0.15, t(22) = 1.34). These temporal decision windows may be linked to the temporal separation of events in episodic memory (Pearson's ρ = -0.53, p = 0.05). Conclusions: Based on our findings, the frequency of visual events accumulated and encoded in the working memory system in the patients' and normal group appears to be approximately (5.7 and 11.2) Hz, respectively. According to the internal clock model, a lower frequency of event pulses tends to result in underestimation of event duration, which phenomenon might be linked to the observed time distortions in patients with dementia.

• 대한핵의학회지
• /
• 제27권2호
• /
• pp.239-247
• /
• 1993

Differentiating the various causes of hydronephrosis from that of obstruction can be very difficult. The decision-making process for those instances of urinary tract dilatation that require surgical correction and those that do not is based in part on the findings of diuresis renography. The methodology for performing this test has differed among nuclear medicine practitioners and the surgical findings are occasionally discrepant from the diuretic renogram interpretation. Consequently we made an automatic computer software program that calculates the slope of the response curve. The quantitative indices, such as the injection and response t1/2 by linear-fitting and monoexponential-fitting, were compared with the visual assessment of the diuretic cinerenography and clinical outcome in 50 children (62 kidneys) with ureteropelvic junction obstruction. Pooled diuresis renogram data indicated that: (1) Visual evaluation of the diuretic cinerenography is a sensitive (87%, 54/62) tool to differentiate obstruction in suspected ureteropelvic junction obstruction. (2) The cut-off value (maximum washout t1/2 with non-obstruction) of injection and response t1/2 by linear-fitting were 40 min. (3) The sensitivity and specificity using injection and response t1/2 by linear-fitting for obstruction were 89%(23/26) and 100%(30/30), respectively. (4) Response t1/2 as well as injection t1/2 by monoexponential-fitting do not stratify children with possible ureteropelivic junction obstruction. In conclusion, quantitative assessment of diuretic renography as well as visual assessment of diuretic cinerenography correlate well with surgical and clinical outcome of suspected ureteropelvic junction obstruction.

• 한국의류학회지
• /
• 제41권2호
• /
• pp.254-265
• /
• 2017

This study examines the possibility of a neuro-scientific approach to fashion Visual Merchandising (VM), by researching the brain activation of customers about fashion stores in terms of VM. Study subjects were in 20's-30's residing in Busan and ten ordinary person or fashion industry related individuals, it measures the change of cerebral blood flow on positive/negative photo stimulus in terms of VM using a functional Near Infrared Spectroscopy (fNIRS) device, and then compared the brain activation to the difference of the fashion store VM. Photo stimuli utilized in the experiment were selected through a preliminary study in advance. The results of this study are as follows. First, the brain activation was found in all 16 channels of stimulus ranges of fashion store VM regardless of positive/negative stimulus. This means that the VM of fashion store causes changes to the cerebral blood flow of consumers, which implies that consumer behavior can be affected by store VM. It also shows that the brain is more active in negative VM stimulus than positive VM despite slight differences in the subjects. In terms of VM, this suggests that the negative factors of fashion stores have a greater effect on the brains of consumers compared to the positive factors. Second, the reaction of the brain channel is different according to the positive/negative VM stimulus of the fashion store by product group and confirms that positive/negative VM stimulus can be distinguished by brain-reaction for the three product groups except for the underwear group among four product groups (men's wear store, women's wear store, underwear store, and sportswear store). The results indicate that more objective scientific measure and decision-making are possible through neuro-science in the strategic execution of VM. This study verified the possibility for a neuro-scientific approach to fashion VM; therefore, there are expectations for the various activation of interdisciplinary research and subsequent development of VM that utilize neuroscience in fashion marketing.