http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
A study on lane path contents for car navigation systems to resolve road inconvenience
Inchul Yang(양인철),Woo Hoon Jeon(전우훈) 한국디지털콘텐츠학회 2019 한국디지털콘텐츠학회논문지 Vol.20 No.1
As shown in many recent studies, the different traffic flow characteristics between lanes are very important in terms of generating and managing traffic information. If avoiding congested road sections is an important purpose of navigation systems, avoiding congested lanes by using traffic state information by lane will also be an important function of navigation systems for resolving road inconvenience in the future. Therefore, this paper proposes a lane-path calculation method required for lane-path services which will be a core content of navigation systems in the future. For this purpose, we proposed a lane-path calculation method for uninterrupted roads by observing the traffic state by each lane. This study collected the travel time by each lane on the Gangbyeon Expressway(Gangbyeonbuk-ro) and analyzed the difference in travel time between lanes. In addition, this study built a road network of the road section and proposed a shortest lane-path calculation method using this, and as a result of comparing the shortest and longest lane-path, there was a significant effect on the reduction of travel time according to the optimal choices of lanes when driving in uninterrupted roads.
양인철(Inchul Yang),전우훈(Woo Hoon Jeon) 한국디지털콘텐츠학회 2019 한국디지털콘텐츠학회논문지 Vol.20 No.7
A various type of road events such as work-zones, accidents, unexpected congestion, and potholes occurs everyday in the road, and it causes a serious traffic congestion and therefore threatens the road users’ safety. Many traffic experts have tried to estimate and predict the traffic conditions under such events, and to develop a traffic model which properly simulate the real world traffic flow change and thus help find a proper traffic operational strategy. Thus a micro traffic simulation model which simulates a temporary lane-close due to the road events was developed in this study. Some previous studies on the traffic simulation model were reviewed and its limitation was found. A new method using plug-in functions provided by the off-the-shelf simulation software program was proposed to tackle the limitation. A traffic model simulating the lane-close was designed, developed, and verified on a toy network. It was demonstrated that the proposed traffic model appropriately represents the real world lane-close phenomenon.
딥러닝 기반 도로위험객체 인식 시스템 성능 향상 방법 개발
양인철(Inchul Yang),전우훈(Woo Hoon Jeon) 한국산학기술학회 2022 한국산학기술학회논문지 Vol.23 No.11
파손된 도로시설물, 로드킬 등의 도로위험객체는 교통사고 위험을 초래하기 때문에 신속하고 정확한 발견과 처리가 매우 중요하다. 이에 전국 도로의 유지관리를 총괄하는 국토교통부에서는 도로이용불편신고 서비스를 운영하고 있으며, 보다 안전하고 효율적인 신고 지원을 위해 이미지 기반의 자동 객체 인식 시스템을 개발 중이다. 시스템 개발은 여러 단계에 걸쳐 수행되고 있으며, 각 단계마다 인식률 향상을 위한 연구가 진행되고 있다. 이에 본 연구에서는 딥러닝 알고리즘을 활용하여 자동으로 도로위험객체를 인식하는 시스템의 성능을 향상시키는 체계적인 방법을 제안하고자 한다. 이를 위해 성능 향상 방법에 대한 절차를 제안하였고, 세부적인 원인 분석을 통해 새로운 솔루션(신경망 구조 변경, 데이터 정제 및 강화)을 제안하였다. 신경망 구조는 기존 YOLOv3보다 우수한 성능을 보이는 YOLOv5로 변경하였고, 데이터 강화를 위해 신고 이미지 중 문제점 확인이 어려운 이미지를 제외하였고, 신고 이미지 외에 오픈소스 이미지 데이터셋에서 양질의 이미지를 추가로 확보하였다. 이를 기존 시스템에 적용하여 신규 시스템을 개발하고, 그 성능을 검증하였다. 본 연구는 기존 시스템에서 인식률이 낮았던 도로시설물과 로드킬을 대상으로 하였으며, 새로운 시스템의 성능 검증을 통해 도로시설물과 로드킬 인식률이 각각 71%에서 91%로, 67%에서 97%로 향상됨을 확인하였다. 향후 제안된 성능 향상 방법에 따라 체계적인 시스템 성능 향상이 가능할 것으로 기대된다. Dangerous road objects cause traffic accidents, which makes it imperative to find and remove them quickly. For this purpose, MOLIT(Ministry of Land, Infrastructure, Transportation) is providing a smartphone app-based reporting service for dangerous road objects and is developing an automatic object recognition and classification system. The development project has several phases, and elaborate efforts have been put into every phase to improve its performance. A systematic method to improve the system performance was proposed in this study. The specific process was proposed, and new solutions(neural network, data manipulation) were found through factor analysis. The existing neural network, YOLOv3, was replaced by the better-performing neural network model, YOLOv5. Some reported images were difficult to designate as problems and were excluded, and high-quality images were also added from an open-source image dataset. The solutions were applied to the system, and its performance was validated. Road facilities and road kills were selected. The results showed that the performance improved the detection rates of road facilities and road kills from 71% to 91% and from 67% to 97%, respectively. The proposed method is expected to improve the system performance systematically.
정밀전자지도 기반의 차로 수준의 위치정보 교환 프레임워크 개발
양인철(Inchul Yang),전우훈(Woo Hoon Jeon) 한국디지털콘텐츠학회 2018 한국디지털콘텐츠학회논문지 Vol.19 No.8
It is necessary to develop a next generation location referencing method with higher accuracy as advanced technologies such as autonomous vehicles require higher accuracy of location data. Thus, we proposed a framework for a lane-level location referencing method (L-LRM) based on high-precision digital road network map, and developed a tool which is capable of analyzing and evaluating the proposed method. Firstly, the necessity and definition of location referencing method was presented, followed by the proposal of an L-LRM framework with a fundamental structure of high-precision digital road network map for the method. Secondly, an architecture of the analysis and evaluation tool was described and then the Windows application program was developed using C/C++ programming language. Finally, we demonstrated the performance of the proposed framework and the application program using two different high precision digital maps with randomly generated road event data.
양인철(Inchul Yang),전우훈(Woo Hoon Jeon) 한국디지털콘텐츠학회 2020 한국디지털콘텐츠학회논문지 Vol.21 No.1
The location based system such as car navigation system and automated driving system matches its location in the road network map using GPS signals on a real-time basis. The basic map matching algorithm, however, runs slow due to its enormous computational complexity and repetitive attribute. Thus the efficient map matching algorithm was proposed using the trajectory of driving vehicles and the topology and direction of road links. Two sub-modules were proposed and included in the algorithm to reduce the number of links for the basic map matching process. The first is Buffered Boundary method, and the second is Driving Direction Comparison method. The performance test results in the Standard-ITS-Link road network demonstrated that the proposed algorithm outperformed the basic map matching algorithm.
양인철(Inchul Yang),전우훈(Woo Hoon Jeon) 한국디지털콘텐츠학회 2018 한국디지털콘텐츠학회논문지 Vol.19 No.11
It is required for traffic congestion relief and safety improvement as well as for self-driving vehicle development to develop an efficient and accurate traffic state observation technology. In addition, the sensing technology is rapidly developing and various types of automotive sensors are adopted to enable ADAS(Advanced Driving Assistant System) in some high-end vehicles. In this regard, we proposed a traffic state observation system using automotive sensors. Widely used automotive sensors were investigated to understand their features, and then some literature on traffic state was reviewed. Six system requirements have been listed, and the system architecture was designed in consideration of the characteristics of sensors and the experts’ opinion. The user interface was carefully designed under assumption of a traffic state observation system with radar, video camera, and GPS. Finally the prototype of the system was developed and tested, and the results show that the proposed system works well.