http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
GPS 수신 정밀도를 고려한 차량 위치 추정 시스템 개발
양찬욱(Chanuk Yang),조정민(Jeongmin Cho),김하영(Hayoung Kim),허건수(Kunsoo Huh) 한국자동차공학회 2017 한국자동차공학회 학술대회 및 전시회 Vol.2017 No.11
Localization is one of the core technology of the autonomous vehicle to improve the performance of autonomous vehicle control. This paper illustrates a vehicle position estimation system considering dilution of precision. The main purpose of this paper is to improve the performance of the pose estimation using GPS received data accuracy called Horizontal Dilution of Precision(HDOP). In this study, we propose a vehicle position estimation algorithm that updates the position states adaptively according to the GPS accuracy, which can be measured as HDOP. This estimation system is validated by vehicle experiment data obtained by the low-priced GPS.
카메라 · 라이다 기반 랜드마크를 활용한 종방향 위치 추정
오하연(Hayeon O),양찬욱(Chanuk Yang),박성준(Seongjun Park),허건수(Kunsoo Huh) 한국자동차공학회 2023 한국자동차공학회 학술대회 및 전시회 Vol.2023 No.11
Localization in autonomous driving is the process by which a vehicle accurately determines its position and orientation. This process includes map matching based on HD maps and provides essential information to the path planning and control systems. It falls under the category of perception within the three major elements of autonomous driving systems: perception, decision-making, and control. Due to the presence of cascading errors or error propagation, where errors from preceding systems can affect the outcomes of subsequent systems, the precision of the perception and decision-making systems, which are the preceding systems, becomes critical. In specific driving environments such as tunnel sections, errors can occur in vehicle position estimation even when using high-precision GPS, due to signal attenuation and reflection within the tunnel. In such situations, lateral localization can be performed based on lane information obtained from the vehicles cameras. However, performing longitudinal localization, which relies on information in the direction of travel, can be challenging due to the lack of relevant data. Therefore, in this paper, we propose a longitudinal localization method for tunnel driving scenarios. We designate tunnel interior structures as landmarks, visually identifiable elements used to determine the vehicles position within its surroundings. We extract information from these landmarks and use it to perform longitudinal localization.
자율주행차 개발 및 평가를 위한 테스트 시나리오 생성 프레임워크
성지훈(Jihoon Sung),김기훈(Gihoon Kim),김웅진(Eungjin Kim),손혁주(Hyukju Shon),양찬욱(Chanuk Yang),최재호(Jaeho Choi),허건수(Kunsoo Huh) 한국자동차공학회 2022 한국자동차공학회 학술대회 및 전시회 Vol.2022 No.11
In general, vehicle development process follows V-Process, which consists of multiple development and evaluation stages, so test scenarios can be one of the ways to set the direction of development. In the case of Advanced Driver Assistance System(ADAS), the evaluation of their performance is conducted in functional unit level, such as lane change assist and lane keeping. However, as the level of the autonomous driving increases, it is necessary to evaluate autonomous driving technology not only at the functional units level, but also whether it is possible to drive within Operational Design Domain(ODD). Until now, for systems higher than level 2, the testing has been replaced by achieving the target distance in the corresponding ODD, but since most of distances are filled on general driving conditions, safety cannot be guaranteed against to various events that may occur in there. Therefore, recently, many studies have been studied to evaluate various situations at the scenario level, such as the PEGASUS project. In this paper, we analyze the already published scenario-level approaches, and construct the relationship between them. Then, an integrated test scenario generation framework is proposed based on organized contents. The proposed method can derive test scenarios for high-level autonomous driving as well as low-level one, and is expected to be used to generate test scenarios for future technology development.