• 한국멀티미디어학회논문지
• /
• 제20권8호
• /
• pp.1271-1281
• /
• 2017

It is essential to collect and analyze target information around the vehicle for autonomous driving of the vehicle. Based on the analysis, environmental information such as location and direction should be analyzed in real time to control the vehicle. In particular, obstruction or cutting of objects in the image must be handled to provide accurate information about the vehicle environment and to facilitate safe operation. In this paper, we propose a method to simultaneously generate 2D and 3D bounding box proposals using LiDAR Edge generated by filtering LiDAR sensor information. We classify the classes of each proposal by connecting them with Region-based Fully-Covolutional Networks (R-FCN), which is an object classifier based on Deep Learning, which uses two-dimensional images as inputs. Each 3D box is rearranged by using the class label and the subcategory information of each class to finally complete the 3D bounding box corresponding to the object. Because 3D bounding boxes are created in 3D space, object information such as space coordinates and object size can be obtained at once, and 2D bounding boxes associated with 3D boxes do not have problems such as occlusion.

• 방송공학회논문지
• /
• 제24권4호
• /
• pp.573-579
• /
• 2019

완전한 자율 주행에 이르기 위해서는 주변 환경을 인지하는 인지 능력이 사람보다 뛰어나야 한다. 자율 주행에서 주로 사용되는 $60^{\circ}$ 협각, $120^{\circ}$ 광각 카메라는 시야각에 따른 각각의 단점이 존재한다. 본 논문의 목적은 광각, 협각 카메라가 가진 각각의 단점을 극복하기 위하여, 다중화각 차량 전방 카메라 시스템을 이용하여 더 넓은 영역의 전방을 대상으로 더 정확히 객체를 인식할 수 있는 심층신경망 알고리즘을 개발하는 것이다. 광각, 협각 카메라로 취득된 데이터의 종횡비를 분석해 SSD(Single Shot Detector) 알고리즘을 수정하였고, 취득된 데이터를 학습하여 단안 카메라만을 사용할 때 보다 높은 성능을 달성하였다.

• 자동차안전학회지
• /
• 제16권2호
• /
• pp.20-26
• /
• 2024

The delivery using drones has been attracting attention because it can innovatively reduce the delivery time from the time of order to completion of delivery compared to the current delivery system, and there have been pilot projects conducted for safe drone delivery. However, the current drone delivery system has the disadvantage of limiting the operational efficiency offered by fully autonomous delivery drones in that drones mainly deliver goods to pre-set landing sites or delivery bases, and the final delivery is still made by humans. In this paper, to overcome these limitations, we propose obstacle detection and landing site selection algorithm based on a vision sensor that enables safe drone landing at the delivery location of the product orderer, and experimentally prove the possibility of station-to-door delivery. The proposed algorithm forms a 3D map of point cloud based on simultaneous localization and mapping (SLAM) technology and presents a grid segmentation technique, allowing drones to stably find a landing site even in places without prior information. We aims to verify the performance of the proposed algorithm through streaming data received from the drone.

• Journal of Biosystems Engineering
• /
• 제37권5호
• /
• pp.287-295
• /
• 2012

Purpose: Some of the most representative approaches are to apply next generation technologies to save energy consumption, fully automated control system to appropriately maintain environmental conditions, and autonomous assistance system to reduce labor load and ensure operator's safety. Nevertheless, improvement of upcoming method for soil cultured greenhouse has not been sufficiently achieved. Geometrical complexity of ground in protected crop cultivation might be one of the most dominant factors in design of autonomous vehicle. While there is a practical solution fairly enough to promise an accurate travelling, such as autonomous sprayer guided by rail or induction coil, for various reasons including the limitation of producer's budget, the previously developed sprayer has not been widely distributed to market. Methods: In this study, we developed an autonomous sprayer considering travelling performance on geometrical complexity of ground in soil cultured greenhouse. To maintain a stable travelling and to acquire a real time feedback, common wire with 80 mm thick and body frame and sprayer boom. To evaluate performance of the prototype, tracking performance, climbing performance and spraying boom's uniform leveling performance were individually evaluated by corresponding experimental tests. Results: The autonomous guidance system was proved to be sufficiently suitable for accurate linear traveling with RMS as lower than approximately 10 cm from designated path. Also the prototype could climb $10^{\circ}$ of ground's slope angle with 40 kg of water weight. Uniform leveling of spraying boom was successfully performed within $0.5^{\circ}$ of sprayer boom's slope. Conclusions: Considering more complex pathways and coarse ground conditions, evaluations and improvements of the prototype should be performed for promising reliability to commercialization.

• 한국기계가공학회지
• /
• 제20권10호
• /
• pp.44-49
• /
• 2021

Seats in autonomous vehicles must be able to rotate to fully utilize the interior space. Generally, ultra-high strength steel is used for the rotation module because it should have high strength and high rigidity. In addition, the rotating parts are difficult to form because they have complex shapes. In this study, the upper plate of the rotating module, whose complex shape makes it difficult to form, was formed by applying the hot stamping method. The drawing method and the form-drawing method, which are generally used to form components of complex shapes, were compared. We showed that the form-drawing method increased the degree of freedom of the material flow to improve the formability, thus enabling the forming of the plate. In addition, the die and blank shapes were found to be important factors in determining the success of the hot stamping. The validity of the analysis results was confirmed through forming analysis and experiments.

• 한국지리정보학회지
• /
• 제24권2호
• /
• pp.102-116
• /
• 2021

4차 산업혁명의 대표라고 할 수 있는 자율주행차량의 안전한 운행을 위해서는 센서 기술, 소프트웨어 기술, 차량 기술 등 다양한 기술 조합이 필요하다. 자율주행차량은 차량 내에 탑재된 다양한 센서를 통해서 현재의 위치정보와 주변 상황을 인지하여 운전자에게 의존하지 않고 스스로 판단하고 주행하는 차량이다. 완전자율주행을 위해서는 완벽한 인지기술이 필요하고 정밀도로지도는 차선, 정지선, 신호등, 횡단보도 등에 대한 정보를 정밀하게 제공하고 있기 때문에 자율주행 차량에서 발생하는 인지 오차를 최소화시킬 수 있음으로, 신뢰성 있는 자율주행차량을 위해서는 도로 위 다양한 시설물들의 위치정보를 차량에 입력한 정밀지도 정보가 필수적이다. 본 연구에서는 정밀도로지도의 정의 및 필요성 국내외 동향을 분석하고 실제 운영되고 있는 대구광역시 자율주행특화지역(수성의료지구, 약 24km)과 세종특별자치시 행복도시(약 33km), 서울대학교 시흥캠퍼스 FMTC(Future Mobility Technical Center) PG(Proving Ground)를 대상으로 국토지리정보원 MMS(Mobile Mapping System) 측량 성과물을 활용하여 정밀도로지도 서비스인 Web GIS 기반 HD(High Definition) Map 프로토타입을 구축하였다. 추후 연구에서는 본 연구에서 구축한 정밀도로지도 서비스를 자율주행차량 및 관제 시스템에 탑재 시켜 실시간 위치검증 및 위치보정 알고리즘의 성능 검증을 진행하고자 한다.

• 문화기술의 융합
• /
• 제3권2호
• /
• pp.49-54
• /
• 2017

한국의 제네시스G80, 티볼리2017등이 부분적 자율주행차 시연을 진행 중이며 상용화에도 박차를 가하고 있다. 세계에서는 테슬라 'X'가 완전자율주행차를 구현하면서 테스트 과정을 완료하였다. 자율주행차가 점차적으로 상용화가 되면서 운전 취약자에 대한 해결책이 제시되고 편의성도 증가하면서 교통사고 비율도 감소할 것으로 예측된다. 반면에 자율주행차 사고 시, 윤리적 책임 문제 등 준비되어야 할 관련 법률은 매우 부족한 실정이며 자율주행차의 보안문제와 개인사생활의 침해도 매우 우려되는 실정이다. 본 논문에서는 현재 자율주행차에 대한 국내외 관련연구와 분석현황을 제시하고, 문제점에 대한 해결책과 활성화 모델을 제안한다.

• 방송공학회논문지
• /
• 제28권3호
• /
• pp.302-313
• /
• 2023

본 연구는 과학기술정보통신부가 2017년부터 1조원 이상을 투자한 'AI Hub 댐' 사업에서 구축된 인공지능 모델 학습데이터의 품질관리를 자동화할 수 있는 프레임워크의 개발을 목표로 한다. 자율주행 개발에 사용되는 AI 모델 학습에는 다량의 고품질의 데이터가 필요하며, 가공된 데이터를 검수자가 데이터 자체의 이상을 검수하고 유효함을 증명하는 데는 여전히 어려움이 있으며 오류가 있는 데이터로 학습된 모델은 실제 상황에서 큰 문제를 야기할 수 있다. 본 논문에서는 이상 데이터를 제거하는 신뢰할 수 있는 데이터셋 정제 프레임워크를 통해 모델의 인식 성능을 향상시키는 전략을 소개한다. 제안하는 방법은 인공지능 학습용 데이터 품질관리 가이드라인의 지표를 기반으로 설계되었다. 한국정보화진흥원의 AI Hub을 통해 공개된 자율주행 데이터셋에 대한 실험을 통해 프레임워크의 유효성을 증명하였고, 이상 데이터가 제거된 신뢰할 수 있는 데이터셋으로 재구축될 수 있음을 확인하였다.

• Ocean Systems Engineering
• /
• 제3권3호
• /
• pp.219-236
• /
• 2013

The accurate prediction of motion in waves of a marine vehicle is essential to assess the maximum sea state vs. operational requirements. This is particularly true for small crafts, such as Autonomous Surface Vessels (ASV). Two different numerical methods to predict motions of a SWATH-ASV are considered: an inviscid strip theory initially developed at MIT for catamarans and then adapted for SWATHs and new a hybrid strip theory, based on the numerical solution of the radiation forces by an unsteady viscous, non-linear free surface flow solver. Motion predictions obtained by the viscous flow method are critically discussed against those obtained by potential flow strip theory. Effects of viscosity are analyzed by comparison of sectional added mass and damping calculated at different frequencies and for different sections, RAOs and motions response in irregular waves at zero speed. Some relevant conclusions can be drawn from this study: influence of viscosity is definitely non negligible for SWATH vessels like the one presented: amplitude of the pitch and heave motions predicted at the resonance frequency differ of 20% respectively and 50%; in this respect, the hybrid method with fully non-linear, viscous free surface calculation of the radiation forces turns out to be a very valuable tool to improve the accuracy of traditional strip theories, without the burden of long computational times requested by fully viscous time domain three dimensional simulations.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제13권9호
• /
• pp.4518-4540
• /
• 2019

With the development of IT convergence technology, autonomous driving has gradually developed; however, the vehicle is still operated by the driver, who should always be in good health - but sometimes, this is not the case. It is especially dangerous to drive when drowsy, and unable to fully concentrate on driving, such as when taking certain medicines, or through fatigue. Drowsy driving is at least eight times more dangerous than normal driving, and as dangerous as drunk driving. Previous research has looked at technology to detect drowsiness, in order to wake up drivers when necessary, or to safely stop the vehicle. Furthermore, many studies have been conducted to find out when drowsiness occurs. However, it is more desirable for the driver to take sufficient rest during a break, in order to be able to continue to focus and drive. In other words, it is important to maintain a normal state before drowsiness. In this study, we introduce a sound source to increase driver concentration and prevent drowsiness, another that can improve the quality of sleep, and a system that produces these sound sources. The proposed system has a noise reduction effect of about 15 dB. We have confirmed that the proposed sound induces an EEG of the desired form.