http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
곽기성(GiSung Gwak),염승재(SeungJae Yum),황성호(Sung-Ho Hwang) 대한기계학회 2014 대한기계학회 춘추학술대회 Vol.2014 No.11
This paper proposes an algorithm to detect the lane on the road using line and lane width information captured by front and AVM camera. First, we perform inverse perspective mapping about image captured by front camera, and then combine front camera image with AVM image. Lastly, we extend region of interest from the front camera image to AVM image to detects the lane because the closer to the vehicle, the higher probability of lane. Experimental results show that the proposed algorithm successfully improved lane detection.
도심 주행을 위한 AVM 영상과 RTK GPS를 이용한 차량의 정밀 위치 추정
곽기성(Gisung Gwak),김동규(DongGyu Kim),황성호(Sung-Ho Hwang) 유공압건설기계학회 2020 드라이브·컨트롤 Vol.17 No.4
To ensure the safety of Advanced Driver Assistance Systems (ADAS) or autonomous vehicles, it is important to recognize the vehicle position, and specifically, the increased accuracy of the lateral position of the vehicle is required. In recent years, the quality of GPS signals has improved a lot and the price has decreased significantly, but extreme urban environments such as tunnels still pose a critical challenge. In this study, we proposed stable and precise lane recognition and tracking methods to solve these two issues via fusion of AVM images and vehicle sensor data using an extended Kalman filter. In addition, the vehicle"s lateral position recognition and the abnormal state of RTK GPS were determined using this approach. The proposed method was validated via actual vehicle experiments in urban areas reporting multipath and signal disconnections.
곽기성(Gisung Gwak),임준영(Junyoung Lim),박찬호(Chanho Park),유동연(Dongyeon Yu),황성호(Sung-Ho Hwang) 대한기계학회 2018 대한기계학회 춘추학술대회 Vol.2018 No.12
Recently, with the rapid development of autonomous driving technology, autonomous driving system has already been commercialized at Level 2. However, existing lane detection technology in Level 2 systems is based on a relatively ideal situation in which the lane is clear as the highway and the curvature is moderate. Therefore, it is difficult to apply and commercialize to complex lane environments such as large curvature roads or off-ramp roads. In this paper, we propose a lane detection algorithm using hough transform algorithm for straight lane detection and RANSAC(Random sample consensus) algorithm. First, the road image is preprocessed using the feature of the road rules of the lane, and then the second order curve fitting is performed by applying the algorithm. Then, in order to detect the correct driving lane in the environment such as off-ramp roads, the information of a global path like navigation information is used. Finally, this algorithm is verified to detect the lane robustly in complex lane environments.
딥러닝 기반 실시간 픽셀 단위 주행 가능 영역 검출 및 격자 지도 생성 알고리즘
김경수(Kyeongsu Kim),곽기성(Gisung Gwak),이재인(Jaein Lee),황성호(Sung-Ho Hwang) 대한기계학회 2021 대한기계학회 춘추학술대회 Vol.2021 No.11
Self-driving, a core technology that leads the Fourth Industrial Revolution era, aims to create a route and driving safely to the target location based on information obtained through various sensors. In real-world driving environments, all processing processes are guaranteed real-time and require high accuracy for different environments. In this paper, we implement a deep learning model that detects drivable areas for real-time autonomous driving and generates noise-removed grid map. The pixel-level segmentation model for detecting drivable area is using BiSeNet v2 with guaranteed real-time and excellent performance. A vehicle simulator was constructed to reflect the vehicle characteristics and various environments of the BRT. Data for deep learning models is used driving videos from vehicle simulator and the Cityscapes dataset. The grid map generation algorithm proposed in this paper eliminates noise for the drivable area, increasing the IoU of the driveable area by up to 3%.
딥러닝 기반 장애물 인식을 위한 가상환경 및 데이터베이스 구축
이재인(JaeIn Lee),곽기성(Gisung Gwak),김경수(KyongSu Kim),강원율(WonYul Kang),신대영(DaeYoung Shin),황성호(Sung-Ho Hwang) 유공압건설기계학회 2021 드라이브·컨트롤 Vol.18 No.4
This study proposes a method for creating learning datasets to recognize obstacles using deep learning algorithms in automated construction machinery or an autonomous vehicle. Recently, many researchers and engineers have developed various recognition algorithms based on deep learning following an increase in computing power. In particular, the image classification technology and image segmentation technology represent deep learning recognition algorithms. They are used to identify obstacles that interfere with the driving situation of an autonomous vehicle. Therefore, various organizations and companies have started distributing open datasets, but there is a remote possibility that they will perfectly match the user"s desired environment. In this study, we created an interface of the virtual simulator such that users can easily create their desired training dataset. In addition, the customized dataset was further advanced by using the RDBMS system, and the recognition rate was improved.