http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
저가형 Camera sensor를 이용한 자율주행 Trailer hitch angle 검출
장성빈(Sung Been Jang),신희석(Heeseok Shin),이제욱(Je Uk Lee) 한국자동차공학회 2020 한국자동차공학회 학술대회 및 전시회 Vol.2020 No.11
With the recent increase in the autonomous vehicle industry, research on self-driving technology has been actively conducted on specially equipped vehicles such as buses and trailers or commercial vehicles, beyond research on autonomous driving based on general passenger vehicles. In particular, the trailer has a tracker part that a person drives and a trailer part that carries luggage. In general, in forward driving, only the length of the vehicle is a little longer, and there is no difficulty in driving. However, when driving in reverse, steering is difficult because the steering angle is opposite to that of a general vehicle. The most important part in reverse driving is the detection of the hitch angle, which is the connection part between the tracker and the trailer. Therefore, this paper aims to detect the trailer hitch angle using the rear camera, which is a low-cost sensor attached to a tracker vehicle, not an encoder or expensive sensor.
실내 환경에서의 트랙터 자동 주차를 위한 카메라 센서 기반의 트레일러 Hitch Joint Angle 검출 알고리즘
장성빈(Sung Been Jang),신희석(Hee seok Shin),김현우(Hyun woo Kim),최윤중(Yoon jung Choi),김정하(Jung Ha Kim) 한국자동차공학회 2022 한국자동차공학회 학술대회 및 전시회 Vol.2022 No.11
Recently, many studies on autonomous vehicles have been conducted, and research on autonomous driving technology has been actively conducted in various fields such as multi-purpose vehicles such as camping cars. In particular, camping cars and large cargo trucks are equipped with trailers on the skin after the vehicle, which are divided into towed trailers and tractors that tow them. The steering angle of the tractor takes precedence over the movement of the trailer because the movement of the tractor and the trailer in the direction of rotation is different when driving backward when parking. Therefore, in order to implement an automatic parking system, we aim to implement an automatic parking algorithm by detecting the angle of the Hitch joint between the tractor and the trailer based on the camera sensor and sending data to determine the steering angle of the tractor.
Deep Learning 기반의 Semantic Segmentation을 활용한 Road Mark 검출
장성빈(Sung Been Jang),정세윤(Se Yun Jung),강동완(Dong Wan Kang),김정하(Jung Ha Kim) 한국자동차공학회 2021 한국자동차공학회 부문종합 학술대회 Vol.2021 No.6
Until recently, research on lane detection and deep learning-based detection algorithms have been actively conducted. When driving straight or changing lanes, the rest of the information except lanes acts as noise, reducing lane recognition accuracy. In particular, the roadmark indicating the direction of progress is similar in colour to the driving lane and thus acts as noise, especially when the vehicle or shadow obscures the triangular area corresponding to the head of the arrow. Therefore, this paper aims to accurately detect road marks by learning and training on Road Mark by using camera sensors to perform deep learning-based Semantic Segmentation.
단안카메라와 YOLO V3 네트워크를 활용한 객체 거리 추정 알고리즘
정세윤(Se Yun Jung),장성빈(Sung Been Jang),백선우(Sun Woo Baek),김정하(Jung Ha Kim) 한국자동차공학회 2021 한국자동차공학회 부문종합 학술대회 Vol.2021 No.6
For self-driving, measuring the distance of forward object is used to several area such as preventing rear-end collisions. Currently, LiDAR in self-drving cars measure distance with high accuracy, but it is difficult to commercialize due to the high cost. In this paper, we propose an algorithm for estimating depth at a low cost using a monocular camera. The algorithm use the video of a monocular camera as input to YOLO V3 Network and use the object detection result data to calculate the pixel unit distance. We define the proportional expression using pixel unit and calculate the depth to object by inverse perspective mapping. The depth estimated by a monocular camera is similar to the distance measured by LiDAR in a specific road section.
NDT를 이용한 실내 Tractor-Trailer 자동 주차 알고리즘 개발
신희석(Heeseok Shin),장성빈(Sung Been Jang),장재익(Jeaik Jang),김현우(Hyunwoo Kim),최윤중(YunJung Choi),김정하(Jung-Ha Kim) 한국자동차공학회 2022 한국자동차공학회 학술대회 및 전시회 Vol.2022 No.11
This paper is about automatically parking indoors through localization using LiDAR in an indoor where GPS is not received. The indoor localization method used the NDT algorithm, and the hitch angle of the trailer was detected using the camera. The position of the tractor was detected using the NDT algorithm, and the position and angle of the trailer were calculated by applying the trailer model based on the hitch angle detected by the camera. The trailer was parked in the designated parking space by following the finally created path.