• Journal of KIISE
• /
• v.44 no.1
• /
• pp.44-50
• /
• 2017

Pedestrian detection, an important component of autonomous driving and driving assistant system, has been extensively studied for many years. In particular, image based pedestrian detection methods such as Hierarchical classifier or HOG and, deep models such as ConvNet are well studied. The evaluation score has increased by the various methods. However, pedestrian detection requires high sensitivity to errors, since small error can lead to life or death problems. Consequently, further reduction in pedestrian detection error rate of autonomous systems is required. We proposed a new method to detect pedestrians and reduce the error rate by using the Faster R-CNN with new developed pedestrian training data sets. Finally, we compared the proposed method with the previous models, in order to show the improvement of our method.

• Journal of Korean Society for Geospatial Information Science
• /
• v.2 no.2 s.4
• /
• pp.81-88
• /
• 1994

The CNS(Car Navigation System) is used more generally in driver aid system than ALV(Auto nomous Land Vehicle) research area. In this paper we developed a new position tracking algorithm for the Global Path Planning in the CNS. In japan, CNS is already well developed and, thesedays they sell CNS products about $400{\sim}500$ thousands per year, and USA and European Communications(EC), too. In Korea, studies of the first generation CNS, which finds current location of a navigating vehicle and displays its location in a Digital-Map with real-time are progressing but still in the beginning step. Therefore a new position tracking algorithm is presented, which reduces vehicle position error dramatically by fusing GPS and dead-reckoning sensors. And the validity of our algorithm is demonstrated by the experimental results with the real car.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.20 no.3
• /
• pp.590-598
• /
• 2016

It comes into the spotlight as the new Blue Ocean in which the connected car industry in which the car and mobile communication technology is convergence. All sorts of infortainments services connecting with the portable electronic device(Smart phone, tablet PC, and MP3 player) and car are rapidly grown. The Connected car emphasizes the vehicle connectivity with the concept that the car has communication with the around on a real time basis and it provides the safety and expedience to the operator and using the thing of Internet (IoT) in the car and supports the application, presently, the entertainment service including the real-time Navigation, parking assistant function, not only the remote vehicle control and management service but also Email, multimedia streaming service, SNS and with the platform. Intelligent vehicle network is studied as the kind according to MANET(Mobile Ad Hoc Network) for the safety operation of the cars of the road and improving the efficiency of the driving.

• Journal of Korean Society of Transportation
• /
• v.18 no.3
• /
• pp.101-112
• /
• 2000

The Purpose of this Paper was to revise the method of USHCM and to establish new method for level of service analysis on two-lane rural highways. For this Purpose, total delay rate was selected as new MOE for level of service, replacing the present Percent time delay. This result showed that total delay rate was more effective for considering the effects of traffic flows, auxiliary lane, and vertical tirade. The application of total delay rate could resolve the Problems in the USHCM method, such as too wide ranges for level of service D and E, and the use of different Procedures for level of service analysis of general terrain segment and specific grade Procedures. The research results are as follows First, a new method for level of service analysis on two-lane rural highways was developed using the total delay rate. Second, a new classification for level of service was developed and a consistent method applicable for general terrain segment and specific tirade Procedures was developed. Third, the desired speed on two-lane rural highways was determined as 85km/h.

• Journal of IKEEE
• /
• v.20 no.4
• /
• pp.351-360
• /
• 2016

Recently, machine learning algorithms, especially deep learning-based algorithms, have been receiving attention due to its high classification performance. Among the algorithms, Convolutional Neural Network(CNN) is known to be efficient for image processing tasks used for Advanced Driver Assistance Systems(ADAS). However, it is difficult to achieve real-time processing for CNN in vehicle embedded software environment due to the repeated operations contained in each layer of CNN. In this paper, we propose a hardware accelerator which enhances the execution time of CNN by parallelizing the repeated operations such as convolution. Xilinx ZC706 evaluation board is used to verify the performance of the proposed accelerator. For $36{\times}36$ input images, the hardware execution time of CNN is 2.812ms in 100MHz clock frequency and shows that our hardware can be executed in real-time.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.6 no.3
• /
• pp.67-76
• /
• 2007

Although Hi-pass have been operated for alleviating traffic congestion and enhancing mobility in expressways since 2000, there is not any standard for safe operation. For that reason, we investigated about tubular markers, which is a typical facility for safety in Hi-pass. Tubular Markers are installed for separating same or opposite traffic flows spatially, supporting the marks at a place which has not only high possibility of accidents but distinguished carefulness of drivers. In this paper, it is noted two considerings; one is prevention of accident from the speed gap of autos and Hi-pass vehicles; the other is guarantee of necessary distance that Hi-pass vehicles could be changes the lanes for off-ramp. Focusing those considerations, it is proposed not only prefer distance of tubular markers, but also fundamental basic subjects. We anticipate that the results of this study would be useful reference in more safe and efficient tollgate operation of Hi-pass.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.51 no.10
• /
• pp.190-196
• /
• 2014

This paper proposes a mobile-optimized pedestrian detection method using Cascade of HOG(Histograms of Oriented Gradients) for ADAS(Advanced Driver Assistance System) on smartphones. In order to use the limited resource of mobile platforms efficiently, the method is implemented by the OpenCL(Open Computing Language) library, and its processing time is reduced in the following two aspects. Firstly, the method sets a program build option specifically and adjusts work group sizes as variety of kernels in the host code. Secondly, it utilizes local memory and a LUT(Look-Up Table) in the kernel code to accelerate the program. For performance evaluation, the developed algorithm is compared with the mobile CPU-based OpenCV(Open Computer Vision) for Android function. The experimental results show that the processing speed is 25% faster than the OpenCV hogcascade.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.15 no.4
• /
• pp.68-80
• /
• 2016

This study proposes the traffic prediction and optimal traffic control system based on cell transmission model and genetic algorithm in cloud environment. The proposed prediction and control system consists of four parts. 1) Data preprocessing module detects and imputes the corrupted data and missing data points. 2) Data-driven traffic prediction module predicts the future traffic state using Multi-level K-Nearest Neighbor (MK-NN) Algorithm with stored historical data in SQL database. 3) Online traffic simulation module simulates the future traffic state in various situations including accident, road work, and extreme weather condition with predicted traffic data by MK-NN. 4) Optimal road control module produces the control strategy for large road network with cell transmission model and genetic algorithm. The results show that proposed system can effectively reduce the Vehicle Hours Traveled upto 60%.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.49 no.4
• /
• pp.85-94
• /
• 2012

Lane marking detection is essential to both ITS and DAS systems. The objective of this paper is to provide more robust technique for lane marking detection than traditional techniques by using scale-space technique. Variable thresholding that is based on the local statistics may be very effective for detecting such objects as lane markings that have prominent intensities. However, such techniques that only rely on local statistics have limitations containing irrelevant areas as well. We reduce the candidate areas by combining the variable thresholding result with cost-efficient horizontal 1D LoG filtered scale space. Through experiments using practical images, we could achieve significant improvement over the techniques based not only on the variable thresholding but also on the Hough transform that is another very popular technique for this purpose.

• Proceedings of the KSR Conference
• /
• 2007.11a
• /
• pp.179-186
• /
• 2007

KTX control systems mutually control OBCS, ATC, MBU, TECA, MDT, ABU, HVAC, TRAE, PID and FDTR, KTX OBCS as master, and controls other sub-control devices as slave, using various serial lines. In order to analyze physical structure of various serial link lines and mutual data transmission structure, serial line analyzer is used in many ways. To use serial line analyzer, prior and professional technics about High Speed Train and experience of using device are necessary. In spite of difficult situation of space and environment where we work for maintenance of High Speed Train, in presenting basic structure about physical connection method aquired by sub-device serial line data collection and about communication data analysis, I hope that this research will be helpful for the person who work for similar area. Also, I hope that this research will help diagnostic work of High Speed Train, which is necessary for test run of independently developed High Speed Train.