Driving Safety Support Systems Research Articles

DETECTION AND LOCALIZATION OF TRAFFIC LIGHTS USING YOLOV3 AND STEREO VISION

Abstract. This paper focus on traffic light distance measurement using stereo camera which is a very important and challenging task in image processing domain, where it is used in several systems such as Driving Safety Support Systems (DSSS), autonomous driving and traffic mobility. In this paper, we propose an integrated traffic light distance measurement system for self-driving based on stereo image processing. Therefore, an algorithm to spatially locate the detected traffic light is required in order to make these detections useful. In this paper, an algorithm to detect, classify the traffic light colours and spatially locate traffic light are integrated. Detection and colours classification are made simultaneously via YOLOv3, using RGB images. 3D traffic light localization is achieved by estimating the distance from the vehicle to the traffic light, by looking at detector 2D bounding boxes and the disparity map generated by stereo camera. Moreover, Gaussian YOLOv3 weights based on KITTI and Berkeley datasets has been replaced with the COCO dataset. Therefore, a detection algorithm that can cope with mislocalizations is required in autonomous driving applications. This paper proposes an integrated method for improving the detection accuracy and traffic lights colours classification while supporting a real-time operation by modelling the bounding box (bbox) of YOLOv3. The obtained results show fair results within 20 meters away from the sensor, while misdetection and classification appeared in further distance.

Distance measurement system for autonomous vehicles using stereo camera

Abstract The focus of this paper is inter-vehicles distance measurement which is a very important and challenging task in image processing domain. Where it is used in several systems such as Driving Safety Support Systems (DSSS), autonomous driving and traffic mobility. In the current paper, we propose an inter-vehicle distance measurement system for self-driving based on image processing. The proposed system uses two cameras mounted as one stereo camera, in the hosting vehicle behind the rear-view mirror. The detection of vehicles is performed first in a single camera using a recent powerful work from the literature. Then, the same vehicle is detected in the image captured by the second camera using template matching technique. Thus, the inter-vehicle distance is calculated using a simple method based on the position of the vehicle in both cameras, geometric derivations and additional technical data such as distance between the cameras and some other specific angles (e.g. the cameras view field angle). The results of the extensive experiments showed the high accuracy of the proposed method compared to the previous works from literature and it allows to measure efficiently the distances between the vehicles and the hosting vehicle. In addition, this method could be used in several systems of various domains in real time regardless of the object types. The experiments results were done on a Hardware Processor System (HPS) located in a VEEK-MT2S provided by TERASIC.

CCN-Based Inter-Vehicle Communication for Efficient Collection of Road and Traffic Information

Recently, inter-vehicle communication, which helps to avoid collision accidents (by driving safety support system) and facilitate self-driving (by dissemination of road and traffic information), has attracted much attention. In this paper, in order to efficiently collect road/traffic information in the request/response manner, first a basic method, Content-centric network (CCN) for Vehicular network (CV), is proposed, which applies CCN cache function to inter-vehicle communication. Content naming and routing, which take vehicle mobility into account, are investigated. On this basis, the CV method is extended (called ECV) to avoid the cache miss problem caused by vehicle movement, and is further enhanced (called ECV+) to more efficiently exploit cache buffer in vehicles, caching content according to a probability decided by a channel usage rate. Extensive evaluations on the network simulator Scenargie, with a realistic open street map, confirm that the CV method and its extensions (ECV, ECV+) effectively reduce the average number of hops of data packets (by up to 47%, 63%, and 83%, respectively) and greatly improve the content acquisition success rate (by up to 356%, 444%, and 689%, respectively), compared to the method without a cache mechanism.

Lane-Change Detection Based on Vehicle-Trajectory Prediction

We propose a new detection method to predict a vehicle's trajectory and use it for detecting lane changes of surrounding vehicles. According to the previous research, more than 90% of the car crashes are caused by human errors, and lane changes are the main factor. Therefore, if a lane change can be detected before a vehicle crosses the centerline, accident rates will decrease. Previously reported detection methods have the problem of frequent false alarms caused by zigzag driving that can result in user distrust in driving safety support systems. Most cases of zigzag driving are caused by the abortion of a lane change due to the presence of adjacent vehicles on the next lane. Our approach reduces false alarms by considering the possibility of a crash with adjacent vehicles by applying trajectory prediction when the target vehicle attempts to change a lane, and it reflects the result of lane-change detection. We used a traffic dataset with more than 500 lane changes and confirmed that the proposed method can considerably improve the detection performance.

1309 大型商用車ドライバの車線逸脱意図の検知(OS6-2 ヒューマンファクタ2,OS6 ヒューマンファクタ,オーガナイズド・セッション(OS))

The installation of various driving safety support systems has been mandatory. However these systems may interfere with driver's operation and this problem leads the driver to distrust the system. If driver's intention is detected accurately, these systems will be more acceptable for the drivers. The purpose of this study is detecting driver's intention of lane departure. The lateral velocity during lane change in driving simulator was analyzed. The results revealed that the lateral velocity was higher in intentional lane departure than unintentional case. In addition, that of lane departing to the right was higher than to the left.

2A2-G06 環境設置型高速ビジョンを用いた高速移動体の衝突回避支援システム(【交通・物流部門】カー・ロボティクスとITS)

In this paper, we propose a driving safety support system (DSSS) which employs high-speed vision installed in surrounding environments such as highways, urban roads, and crossroads. The system is aiming to recognize traffic situations including hidden places from mobilities and utilize such information for driving safety support. The system consists of synchronized high-speed cameras which enable us to acquire images every millisecond, and mobilities capable of communicating with these high-speed cameras via communication stations. We conducted experiments on collision avoidance and demonstrate that the proposed system can resolve the deficiency of readiness, which is common to environmental vision systems, by introducing high-speed vision.

J0120101 交通事故シミュレーションと安全運転支援システムの評価 : 交通事故シミュレーションの活用とマネジメント その1([J012]交通の安全・環境シミュレーション)

J0120101 交通事故シミュレーションと安全運転支援システムの評価 : 交通事故シミュレーションの活用とマネジメント その1([J012]交通の安全・環境シミュレーション)

Dromosagnosia, or why some people lose their sense of direction while driving

We coined a new word, “dromosagnosia”, from the Greek words, dromos (“way, road”) + agnosia, to describe the loss of direction while driving, an orientation disorder similar to but different from pure topographic disorientation. Historically, human beings have moved more quickly, from using domesticated animals to high speed vehicles, and this may be beyond the brain’s ability to react. Without the benefit of an automatic navigation system, automobiles are associated with more problems of dromosagnosia than are fast-moving aircraft or ships. Previous studies have noted that some areas of the brain are associated with spatial orientation, spatial memory, and even emotion, and abnormalities there could exacerbate the loss of sense of direction. We hypothesize that some people are especially disadvantaged from these brain differences and emotional disturbances when driving their cars. Functional magnetic resonance imaging (fMRI) and event-related potentials (ERP) studies combined with a virtual reality driving simulation might be used to find the areas of the brain related to dromosagnosia. Future applications: some people with dromosagnosia might benefit from special remedial training and a driving safety support system to avoid potential problems.

Real-Time Vehicle Detection using a Single Rear Camera for a Blind Spot Warning System

This paper describes a vision-based vehicle detection system for a blind spot warning function. This detection system has been designed to provide ample performance as a driving safety support system, while streamlining the image processing algorithm so that it can be processed using the computational power of an existing ECU. The procedure used by the system to detect a vehicle in a blind spot is as follows. The system consists of four functional components: obstacle detection, velocity estimation, vertical edge detection, and final classification. In obstacle detection, a predicted image is generated under the assumption that the road surface is a perfectly flat plane, and then an object is detected based on a histogram that is created by comparing the predicted image and an actually observed image. The velocity of the object is estimated by tracking the histogram over time, assuming that both the object and the host vehicle are traveling in the same direction. Vertical edge detection is employed so as to avoid misdetection due to a vehicle shadow projected onto the road surface. In final classification, a vehicle is detected on the basis of these results. The effectiveness of the system was verified by conducting road tests on highways and narrow streets with two-way traffic. Language: en

An Empirical Study of Measures for Preventing Crossing Path Collisions at Unsignalized Intersections—Development of Driving Safety Support Systems using Infrastructure-Vehicle Communication

Measures for preventing the very occurrence of traffic accidents are being viewed with increasing importance given the continued high levels of accidents. This study focused on the fact that the greater majority of traffic accidents occur because of drivers’ perception error concerning risk factors. A study was made of measures for preventing crossing path collisions at unsignalized intersections from the standpoint of improving drivers’ perception abilities. It was hypothesized that calling drivers’ attention to risk factors would help to avoid perception error and work to prevent accidents. Toward that end, two driving safety support systems are proposed that work in cooperation with the roadside communications infrastructure. Two systems can obtain and present to the driver as quickly as possible information on risk factors that drivers are apt to overlook. If necessary, the systems issue audible and visible alerts through the car navigation system to call the driver’s attention to a potential risk. The utility of these support systems as active safety measures was validated in a field operational test involving participation by 2,000 ordinary drivers.

1A2-B21 モービルマッピングシステムを用いた次世代デジタル道路地図作成システムに関する研究

In these years, various research and development for achievement of the driving safety support systems (DSSS) in next generation Intelligent Transport Systems (ITS) are increasing. At the same time, more detail and various road data like road alignments, road gradients, intersection detail data and lane detail data are demanded for the achievement of next generation ITS. But these road data are not recorded in the present digital road map (DRM) which is used in car navigation system. In addition, aerial survey which is one of the mainstream methods of measuring DRM data is not the ideal method on the point of cost and efficiency. A Mobile Mapping System (MMS) that is a vehicle consists of GPS/IMU combined navigation system, cameras, laser scanners, is one of the methods to measure detail road data with high precision. This paper describes the algorithm for measuring detail and various road data for creating next generation DRM using MMS.

1A2-C10 安全運転支援システムのための移動障害物に対する危険度評価方法

1A2-C10 安全運転支援システムのための移動障害物に対する危険度評価方法