• 대한전기학회논문지:시스템및제어부문D
• /
• 제53권10호
• /
• pp.703-713
• /
• 2004

This paper presents the new obstacle avoidance method that is composed of vision and sonar sensors, also a navigation algorithm is proposed. Sonar sensors provide poor information because the angular resolution of each sonar sensor is not exact. So they are not suitable to detect relative direction of obstacles. In addition, it is not easy to detect the obstacle by vision sensors because of an image disturbance. In This paper, the new obstacle direction measurement method that is composed of sonar sensors for exact distance information and vision sensors for abundance information. The modified splitting/merging algorithm is proposed, and it is robuster for an image disturbance than the edge detecting algorithm, and it is efficient for grouping of the obstacle. In order to verify our proposed algorithm, we compare the proposed algorithm with the edge detecting algorithm via experiments. The direction of obstacle and the relative distance are used for the inputs of the fuzzy controller. We design the angular velocity controllers for obstacle avoidance and for navigation to center in corridor, respectively. In order to verify stability and effectiveness of our proposed method, it is apply to a vision and sonar based mobile robot navigation system.

• 한국정보통신학회:학술대회논문집
• /
• 한국해양정보통신학회 2003년도 추계종합학술대회
• /
• pp.264-267
• /
• 2003

본 논문은 DSP를 이용한 스테레오 비젼 로봇의 설계에 관한 연구이다. 스테레오 비젼 로봇은 제어 시스템, 비젼 시스템, 그리고 호스트 컴퓨터로 구성된다. 비젼 시스템은 32비트 DSP 프로세서를 기반으로 구현하였고, 스테레오 영상 처리는 상관계수법을 적용하였다. 실험 결과, 영상인식에 의해 로봇의 제어가 원활하게 되었으며, 영상인식률은 약 95%를 얻었다.

• 전자공학회논문지B
• /
• 제29B권9호
• /
• pp.37-49
• /
• 1992

The standard definition of computational vision is a set of inverse problems of recovering surfaces from images. Thus the common characteristics of the most early vision problems are ill-posed. The main idea for solving ill-posed problems is to restrict the class of admissible solutions by introducing suitable a priori knowledge. Standard regurarization methods lead to satisfactory solutions of early vision problems but cannot deal effectively and directly with a few general problems, such as discontinuity and fusion of information from multiple modules. In this paper, we discuss limitations of standard regularization theory and present new stochastic method. We will outline a rigorous approach to overcome part of ill-posedness of image restoration, edge detection, and stereo vision problems, based on Bayes estimation and MRF(Markov random field) model, that effectively deals with the problems. This result makes one hope that this framework could be useful in the solution of other vision problems.

• Agricultural and Biosystems Engineering
• /
• 제3권1호
• /
• pp.35-43
• /
• 2002

The spectral color resolution of an image is very important in color image analysis. Two factors influencing the spectral color resolution of an image are illumination intensity and lens aperture for a selected vision system. An optimal combination of illumination intensity and lens aperture for color image analysis was determined in the study. The method was based on a model of dynamic range defined as the absolute difference between digital values of selected foreground and background color in the image. The role of illumination intensity in machine vision was also described and a computer program for simulating the optimal combination of two factors was implemented for verifying the related algorithm. It was possible to estimate the non-saturating range of the illumination intensity (input voltage in the study) and the lens aperture by using a model of dynamic range. The method provided an optimal combination of the illumination intensity and the lens aperture, maximizing the color resolution between colors of interest in color analysis, and the estimated color resolution at the combination for a given vision system configuration.

• 한국정밀공학회:학술대회논문집
• /
• 한국정밀공학회 1997년도 추계학술대회 논문집
• /
• pp.500-503
• /
• 1997

In this paper, a method for recognizing 3-D objects using the 3-D Hough transform and the robust stereo vision is studied. A 3-D object is recognized through two steps; modeling step and matching step. In modeling step, features of the object are extracted by analyzing the IGES file. In matching step, the values of the sensed image are compared with those of the IGES file which is assumed to location and orientation in the 3-D Hough transform domain. Since we use the 3-D Hough transform domain of the input image directly, the sensitivity to the noise and the high computational complexity could be significantly allcv~ated. Also, the cost efficiency is improved using the robust stereo vision for obtaining depth map image which is needed for 3-D Hough transform. In order lo verify the proposed method, real telephone model is recognized. Thc results of the location and orientation of the model are presented.

• Journal of Welding and Joining
• /
• 제17권2호
• /
• pp.53-60
• /
• 1999

Conventionally, CCD camera and vision sensor using the projected pattern of light is generally used to inspect the weld bead defects. But with this method, a lot of time is needed for image preprocessing, stripe extraction and thinning, etc. In this study, laser vision sensor using the scanning beam of light is used to shorten the time required for image preprocessing. The software for deciding whether the weld bead is in proper shape or not in real time is developed. The criteria are based upon the classification of imperfections in metallic fusion welds(ISO 6520) and limits for imperfections(ISO 5817).

• 센서학회지
• /
• 제20권3호
• /
• pp.178-186
• /
• 2011

Line tracking is a well defined method of mobile robot navigation. It is simple in concept, technically easy to implement, and already employed in many industrial sites. Among several different line tracking methods, magnetic sensing is widely used in practice. In comparison, vision-based tracking is less popular due mainly to its sensitivity to surrounding conditions such as brightness and floor characteristics although vision is the most powerful robotic sensing capability. In this paper, a vision-based robust path line detection technique is proposed for the navigation of a mobile robot assuming uncontrollable surrounding conditions. The technique proposed has four processing steps; color space transformation, pixel-level line sensing, block-level line sensing, and robot navigation control. This technique effectively uses hue and saturation color values in the line sensing so to be insensitive to the brightness variation. Line finding in block-level makes not only the technique immune from the error of line pixel detection but also the robot control easy. The proposed technique was tested with a real mobile robot and proved its effectiveness.

• Smart Structures and Systems
• /
• 제29권5호
• /
• pp.715-728
• /
• 2022

Recently, numbers of long span pedestrian suspension bridges have been constructed worldwide. While recent tragedies regarding pedestrian suspension bridges have shown how these bridges can wreak havoc on the society, there are no specific guidelines for construction standards nor safety inspections yet. Therefore, a structural health monitoring system that could help ensure the safety of pedestrian suspension bridges are needed. System identification is one of the popular applications for structural health monitoring method, which estimates the dynamic system. Most of the system identification methods for bridges are currently adapting output-only system identification method, which assumes the dynamic load to be a white noise due to the difficulty of measuring the dynamic load. In the case of pedestrian suspension bridges, the pedestrian load is within specific frequency range, resulting in large errors when using the output-only system identification method. Therefore, this study aims to develop a system identification method for pedestrian suspension bridges considering both input and output of the dynamic system. This study estimates the location and the magnitude of the pedestrian load, as well as the dynamic response of the pedestrian bridges by utilizing artificial intelligence and computer vision techniques. A simulation-based validation test was conducted to verify the performance of the proposed system. The proposed method is expected to improve the accuracy and the efficiency of the current inspection and monitoring systems for pedestrian suspension bridges.

• 대한임베디드공학회논문지
• /
• 제19권2호
• /
• pp.73-81
• /
• 2024

Recently, mobile manipulators have been utilized in agriculture industry for weed removal and harvest automation. This paper proposes a domain adaptive fruit detection method for harvest automation, by utilizing OWL-ViT model which is an open-vocabulary object detection model. The vision-language model can detect objects based on text prompt, and therefore, it can be extended to detect objects of undefined categories. In the development of deep learning models for real-world problems, constructing a large-scale labeled dataset is a time-consuming task and heavily relies on human effort. To reduce the labor-intensive workload, we utilized a large-scale public dataset as a source domain data and employed a domain adaptation method. Adversarial learning was conducted between a domain discriminator and feature extractor to reduce the gap between the distribution of feature vectors from the source domain and our target domain data. We collected a target domain dataset in a real-like environment and conducted experiments to demonstrate the effectiveness of the proposed method. In experiments, the domain adaptation method improved the AP50 metric from 38.88% to 78.59% for detecting objects within the range of 2m, and we achieved 81.7% of manipulation success rate.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2004년도 ICCAS
• /
• pp.1920-1925
• /
• 2004

This paper describes a new method for 3D object recognition which used surface segment-based stereo vision. The position and orientation of an objects is identified accurately enabling a robot to pick up, even though the objects are multiple and partially occluded. The stereo vision is used to get the 3D information as 3D sensing, and CAD model with its post processing is used for building models. Matching is initially performed using the model and object features, and calculate roughly the object's position and orientation. Though the fine adjustment step, the accuracy of the position and orientation are improved.