Archives Search

Abstract

We introduce "Road Alignment Data" which has reduced the weight of the map base data required for automated driving technology. We digitize the road alignment data, its changing point and the installation position of signs etc. (coordinates) from the management drawing of...Lead moreWe introduce "Road Alignment Data" which has reduced the weight of the map base data required for automated driving technology. We digitize the road alignment data, its changing point and the installation position of signs etc. (coordinates) from the management drawing of the expressway owned by the expressway company. We shoot the signs etc. (landmarks) by the in-vehicle camera, and specify the position of the vehicle. And the read-ahead road alignment information is acquired based on the position information from the road alignment database mounted on the vehicle. This is digitized based on the existing drawing information and it is not necessary to perform on-site measurement, so it can be created at low cost, the amount of data is very compact, and it can be used for road-to-vehicle communication (ETC 2.0) etc. Update is possible.

Abstract

In this paper the pedestrian positioning system in the intersection using UWB(Ultra-Wide Band) is studied. In this system, UWB signal is transmitted from the terminal owned by each pedestrian and it is received by base stations attached to traffic lights for pedestrian, a...Lead moreIn this paper the pedestrian positioning system in the intersection using UWB(Ultra-Wide Band) is studied. In this system, UWB signal is transmitted from the terminal owned by each pedestrian and it is received by base stations attached to traffic lights for pedestrian, and the distance to the pedestrian is measured. Distance between the pedestrian and the base station is estimated by TOA(Time Of Arrival) of the received signal. The position of the pedestrian is estimated by LSM(Least Square Method) using the distance measurement value. The accuracy of UWB positioning system deteriorates due to thermal noise and multipath. We use the Kalman filter to estimate the position with high accuracy from observed values including errors. In addition, by estimating the walking speed of a pedestrian, it is possible to estimate the future position of a pedestrian. As the results of computer simulations, it is shown that the proposed method achieves high accuracy positioning performance.

Abstract

Singapore has an electronic smart card ticketing system for public transport. The spatial-temporal transaction data from this electronic ticketing system has been used by many researches to study public transport commuter flow and volume. They have highlighted the potenti...Lead moreSingapore has an electronic smart card ticketing system for public transport. The spatial-temporal transaction data from this electronic ticketing system has been used by many researches to study public transport commuter flow and volume. They have highlighted the potential for this to be used for supporting public transport operations and planning purposes. Some have also developed methods using it to analyse home-work patterns, with the potential to inform land use planning. This paper pushes the frontier with new practical methodologies and applications of the data for planning of physical infrastructure to support active modes of commute (e.g. cycling and walking) where there are limited insights on commuters’ experience and preferences for these modes. It will also share from a practice perspective the lessons gained when applied in a real-world setting.

Abstract

In this paper, we propose a new frame of image geo-locating based on structured street view with GPS tag. Our methodology is to find certain neighbor features and determine the matching rate for each image in multiple nearest neighbors in order to sort all candidate image...Lead moreIn this paper, we propose a new frame of image geo-locating based on structured street view with GPS tag. Our methodology is to find certain neighbor features and determine the matching rate for each image in multiple nearest neighbors in order to sort all candidate images globally. First of all, we extract the local features(SIFT) from query image as query features; Secondly, we retrieve K nearest neighbors and prune the indistinctive features for each query feature; Then we take the reference images corresponding to the nearest neighbors as the candidate images and select the feature in multiple nearest neighbors within certain range; Finally, we select the candidate image that has the largest amount of feature points and the highest matching rate as the best match. We evaluated the proposed image geo-localization framework on a 102k public dataset, and it outperforms the existing method by 1% within 180m.

Abstract

This paper documents the challenges met and strategies employed by Singapore’s Land Transport Authority (LTA) concerning operational safety at signalised intersections of close proximity. Due to the highly dense road network and city landscape in Singapore, there is an in...Lead moreThis paper documents the challenges met and strategies employed by Singapore’s Land Transport Authority (LTA) concerning operational safety at signalised intersections of close proximity. Due to the highly dense road network and city landscape in Singapore, there is an inherent need to address motorist behaviour when na vigating intersections of close proximity considering the residue of motorists who may face difficulty in interpreting traffic light signals. In response to potential risks in driver behaviour at such intersections, LTA has adopted strategies, such as modifications to phasing sequence and/or to traffic light infrastructure. This paper subsequently discusses and evaluates multi-faceted approaches in two case studies to create safer road networks and enrich travel experience in Singapore’s people-centred land transport system.

Abstract

The spread of connected vehicles equipped with DSRC/IEEE 802.11p technology is progressing. These connected vehicles broadcast V2X data such as position and speed of the own to the surroundings for the purpose of safe driving support to prevent collision. This V2X data al...Lead moreThe spread of connected vehicles equipped with DSRC/IEEE 802.11p technology is progressing. These connected vehicles broadcast V2X data such as position and speed of the own to the surroundings for the purpose of safe driving support to prevent collision. This V2X data also has potential as a new traffic information source for traffic managements such as traffic signal control. Therefore, we have developed a technology to generate traffic information by fusing V2X data with infrastructure data collected from vehicle detectors, which can be reduced to the minimum number. By using this technology, it is expected to realize a sustainable and lower cost cooperative traffic signal control system. This paper shows the outline of the developed technology, the performance evaluation results of the cooperative traffic signal control by the simulation model, and the results of the queue estimation accuracy verification experiment in real field.

Abstract

The NZTA is in the process of establishing a nationally consistent framework that enables seamless integration of Transport Technology into the existing operational environment with reusable tools and templates, repeatable processes and a consistent management approach fo...Lead moreThe NZTA is in the process of establishing a nationally consistent framework that enables seamless integration of Transport Technology into the existing operational environment with reusable tools and templates, repeatable processes and a consistent management approach for Transport Technology projects and capital project with Transport Technology components. This paper provides the relevant background context followed by an overview of the methodology undertaken, the delivered outcomes and key lessons learnt.

Abstract

Intelligent Transport System (ITS) technologies have enabled efficient monitoring and management of traffic situations on the roads. Through the processing of measurement from fixed and mobile sensors, real-time traffic conditions are made available to traffic operators i...Lead moreIntelligent Transport System (ITS) technologies have enabled efficient monitoring and management of traffic situations on the roads. Through the processing of measurement from fixed and mobile sensors, real-time traffic conditions are made available to traffic operators in the traffic operations control centre (OCC). However, further improvements in incident response require not only real-time situational awareness but also anticipatory decision support system, especially when multiple incidents of varying severity happen at the same time. The Land Transport Authority (LTA) of Singapore aims to develop an integrated real-time traffic simulation and prediction system (i-SPRINT) that connect with its central ITS hub that collects all the traffic data and information from the various ITS sensors deployed around Singapore roads. The i-SPRINT is envisaged to be a highly integrated tool that offers the different levels of traffic simulation from macroscopic to mesoscopic, to microscopic for all the transport modelling needs in Land Transport Authority of Singapore (LTA), both in offline and 24/7 real-time modes for traffic operators and planners. This paper details the process in developing i-SPRINT and the technical features of the system.

Abstract

Transport policy is rightly concerned with reducing road deaths and serious injuries. A challenge it faces is that addressing risk in known areas of harm often just pushes the issue down the road. Traditional analysis has not had the data to try and get ahead of the probl...Lead moreTransport policy is rightly concerned with reducing road deaths and serious injuries. A challenge it faces is that addressing risk in known areas of harm often just pushes the issue down the road. Traditional analysis has not had the data to try and get ahead of the problem by assessing risk across the whole network. This exploratory data analysis uses probe data from EROAD, a global regulatory telematics company, which has installed devices in over 65,000 light and heavy vehicles. This study was conducted on over a billion gps events that describe the activity of a single month. The study aims demonstrate how this data can be used to understand the dynamic risk on the road, and provide a framework for route choice that is based around the accumulated dynamic risk caused by other road users. This suggests a new level of targeted safety management is possible.

Abstract

Research & development, and demonstration experiments related autonomous vehicles are very active in the world for last few years. The basic concept of autonomous vehicles is to drive automatically by sensing and judging using own in-vehicle’s sensors. However, there are ...Lead moreResearch & development, and demonstration experiments related autonomous vehicles are very active in the world for last few years. The basic concept of autonomous vehicles is to drive automatically by sensing and judging using own in-vehicle’s sensors. However, there are some challenging points at junction, railway crossing and merging point of expressway, because the detection range of the in-vehicle’s sensors is limited. Therefore, some infrastructure to support by notification of the necessary information to autonomous vehicles is expected to be developed and implementation. As for Japanese metropolitan highway, there are a lot of merging points whose ramp are very short and not enough distance to accelerate to the same speed as vehicles running at main highway line, because it is almost constructed on ordinary road as viaduct due to space constrain. This paper describes about approaching for realization of merging point support system as one of the above infrastructure.

Abstract

Toward an automated-driving system, we have developed “Infrastructure Radar System” based on 79GHz band, and consider the possibility of V2I application as roadside sensors. The system needs to function properly even under adverse environmental conditions such as meteorol...Lead moreToward an automated-driving system, we have developed “Infrastructure Radar System” based on 79GHz band, and consider the possibility of V2I application as roadside sensors. The system needs to function properly even under adverse environmental conditions such as meteorological situations. This paper presents basically experiment under meteorological situations assumed snowy weather using the infrastructure radar system. Firstly, it was clarified the snow attenuation characteristics under various snowy situations by snowfall simulator. And then, it was verified the effectiveness of practical performance under actual snowy weather in outdoor field.

Abstract

In most countries, drivers are penalized for traffic infringements to prevent recurrences. However, unless a precise understanding of the effect of the penalty is preceded, some penalties for violations may not improve driving habits due to the weak deterrent effect. Conv...Lead moreIn most countries, drivers are penalized for traffic infringements to prevent recurrences. However, unless a precise understanding of the effect of the penalty is preceded, some penalties for violations may not improve driving habits due to the weak deterrent effect. Conversely, the punishments for some other traffic violations can be excessive. In this study, we analysed the effect of various punishments for traffic violations by survival analysis based on data that tracked new drivers for five years from June 2012 to June 2017 in the Republic of Korea. The time-varying Cox proportional hazard model was used to examine change in drivers’ risk of infringement over time. As a result, we found that several punishments did not improve the drivers’ driving behaviours, and, even if the punishments had specific deterrent effects, these effects diminish and disappear after a certain period of time. These results suggest that it is difficult to prevent recurrences for a long period of time because the general deterrence effect of the punishment is not enough to reduce infringements. This finding suggests that additional penalties for drivers who have additional infringements within a certain period of time are effective in improving these drivers’ compliance with the traffic laws.

Abstract

To restrict vehicle ownership and road usage are the main Travel Demand Management (TDM) policies implemented in China metropolises, however, the implementation effect remains controversial. In this article, we combine traditional boundary line survey with modern license-...Lead moreTo restrict vehicle ownership and road usage are the main Travel Demand Management (TDM) policies implemented in China metropolises, however, the implementation effect remains controversial. In this article, we combine traditional boundary line survey with modern license-plate recognition (LPR) technology, and have a clear vision of vehicle operation features changes from vehicle level. Metropolis Shenzhen takes an ideal sample for our study. In detail, several vehicle operation features are chosen, such as vehicle frequency, the characteristics of commuting-like vehicle and so on. The finding suggested that the TDM policies are effective in curbing rapid traffic growth and gaining time for city’s public transportation development. In the short term, the Average Travel Rate for vehicle in central Shenzhen has fallen by 5.5%, Commute-like Vehicle Contribution in Peak-hours has also decreased by 2.9%, by comparison, the indexes for suburbs where policies doesn’t take effect, has increased by 15% and 16% respectively.

Abstract

Before an autonomous vehicle (AV) is allowed to ply on public roads, the vehicle has to be subjected to rigorous testing to show that it is safe for all road-users under various real-life situations. Testing involves assessment of the AV’s driving behaviour during numerou...Lead moreBefore an autonomous vehicle (AV) is allowed to ply on public roads, the vehicle has to be subjected to rigorous testing to show that it is safe for all road-users under various real-life situations. Testing involves assessment of the AV’s driving behaviour during numerous test trips. The assessment would involve reviewing of logged data and videos throughout the test trips. An automated system is presented in this paper that can help to identify and classify various scenarios the AV has gone through during the test trips. This would in effect increase the productivity of the testing, assessment and incident investigation process. The system was built and tested with an AV at the Singapore one-north AV trial testbed and showed that it can identify, classify and assess the AV’s behaviours effectively. This paper describes the design of the system and the tests that the system was put through.

Abstract

Colour is well established as a mechanism to provide simple indication of motorway congestion levels. In Auckland we wondered if greater benefit could be derived by displaying estimated journey times which used colour to compare today’s journey with the ‘normal’ journey t...Lead moreColour is well established as a mechanism to provide simple indication of motorway congestion levels. In Auckland we wondered if greater benefit could be derived by displaying estimated journey times which used colour to compare today’s journey with the ‘normal’ journey time historically on that day of the week and time of day. We found that people intuitively associate red, amber and green colours on VMS with traffic signal commands. Our challenge was to deploy a solution which could facilitate in-trip decision making but was not ‘self-explaining’.

Abstract

To build a sustainable and clean environment, many governments advocate the use of public transit to reduce pollution which arises due to road congestion. In South Korea, efforts to make transit services more popular include the introduction of integrated transport system...Lead moreTo build a sustainable and clean environment, many governments advocate the use of public transit to reduce pollution which arises due to road congestion. In South Korea, efforts to make transit services more popular include the introduction of integrated transport systems which encourages transfers between buses and different transport modes. However, it is difficult to develop such strategies without knowing the factors affecting transit ridership and its sustainability. Previous studies focused on subway ridership estimation using sustainable transport-inducing factors as land-use patterns. There is a limitation in research capturing the endogeneity of bus ridership inherent in subway ridership models. Therefore, the goal of this study is to develop a more plausible model of subway ridership which captures unobserved bias. Secondly, the study aims to find the relationship between subway ridership and ridership inducing variables focusing on land use near subway stations. We collected potential explanatory variables affecting bus ridership and the effects of retail activities within a 500-meter radius around 248 subway stations. Using a two-stage least squares model, bus ridership variable was significant which means it is essential to check endogeneity. It was identified that land use activities near the subway stations had a great impact on subway ridership.

Abstract

Visual impaired people are always suffered from social exclusion because of unfriendly public transit system, especially bus service. They are difficult to understand when the desired bus service arrives and if the approaching bus is their request for or not. Without supp...Lead moreVisual impaired people are always suffered from social exclusion because of unfriendly public transit system, especially bus service. They are difficult to understand when the desired bus service arrives and if the approaching bus is their request for or not. Without supplement facility, the visual impairment could not ‘see’ bus real-time information to understand bus arrival time from smartphone app and LED display on bus stop, and the approaching bus route number. Taichung City government has developed a bus service auxiliary system for visual impaired people. The system is APP-oriented. Visual impaired people can book bus service when they reach the bus stops. The bus operation center will receive the booking and send a message to the designated bus driver. The APP will provide an innovative interactive audio solution in real-time to inform passengers the approaching bus route number and destination. The bus driver will be reminded to serve the blind and visual impaired passenger. In addition, the service will be created by open protocol which means the range of opportunities for future development is unlimited.

Abstract

MaaS (Mobility as a Service) has been seen as the new transportation paradigm. Two MaaS projects were initiated in northern and southern Taiwan in 2017, and UMAJI and MENGO apps have been developed by the projects. This paper presents how the two different MaaS app – UMAJ...Lead moreMaaS (Mobility as a Service) has been seen as the new transportation paradigm. Two MaaS projects were initiated in northern and southern Taiwan in 2017, and UMAJI and MENGO apps have been developed by the projects. This paper presents how the two different MaaS app – UMAJI and MENGO – have been developed depending the contexts of land use and transport contexts of the service areas. In addition, this paper shows the functions and future development of the two MaaS app. UMAJI app is focusing on intercity traveler. Journey planner and online payment for public transit tickets are the basic functions for UMAJI app. Private vehicle users and public transit users are all the target customers for UMAJI app in order to trigger behavior change from private vehicle to public transit and shared modes. Hence, predict travel time, better departure time, better travel routes and enroute re-planning due to traffic events are the functions for private vehicle users. AI(Artificial Intelligent) behavior analysis is the major engine behind the UMAJI app in order to understand our users. As for the MENGO app, Monthly package which integrate bus, metro, ferry, public bike, is the main function.

Abstract

High traffic-related injuries and deaths cause a large amount of social cost in Taiwan. In 2017, traffic accidents took 93 lives in Taichung. Among them, near fifty percent were motorcyclist due to the very high proportion of motorcycle usage. Many young lives and familie...Lead moreHigh traffic-related injuries and deaths cause a large amount of social cost in Taiwan. In 2017, traffic accidents took 93 lives in Taichung. Among them, near fifty percent were motorcyclist due to the very high proportion of motorcycle usage. Many young lives and families have been severely affected by traffic accidents. In order to tackle the safety problem, Taichung city has established the smart cooperative intersection collision avoidance system. This system can detect and collect road traffic conditions through radar and road-side unit. It will calculate time to collision between every two vehicles approaching the intersection. If there is a high crash risk event, an alert will be triggered and display through CMS on the road to provide the driver warning message. This system can be further combined with self-driving vehicles to provide SPAT(Signal Phase and Timing), TSP(Transit Signal Priority) and roadside blind spot sensing ability, to assist self-driving vehicles to pass the intersection. Two intersections are chosen to demonstrate in Taichung City. One is the controlled intersection and the other is the open intersection. The result shows that the system can effectively reduce traffic accidents at intersections.

Abstract

This paper presents a study of rear view monitor system for motorcycles. This system shows the backward image with a camera attached to the helmet and reduces the blind area by displaying it. We use the camera module of Raspberry Pi for camera and Android device to displa...Lead moreThis paper presents a study of rear view monitor system for motorcycles. This system shows the backward image with a camera attached to the helmet and reduces the blind area by displaying it. We use the camera module of Raspberry Pi for camera and Android device to display the video, and they are wirelessly connected by Wi-Fi Direct. In addition, we investigated how much the blind area was with only the left and right mirrors, and investigated how much the blind area could be reduced by rear view monitor system. As a result, by using this system it became possible to check the range 60°behind the motorcycle.

Abstract

This article presents the safety and health status of Meru-Menora Tunnel by monitoring harmful gases emitted from traffic emission which could lead to fire. SICK sensor is used to measure harmful gases concentration comprising of Nitrogen Dioxide (NO2) and Carbon Monoxide...Lead moreThis article presents the safety and health status of Meru-Menora Tunnel by monitoring harmful gases emitted from traffic emission which could lead to fire. SICK sensor is used to measure harmful gases concentration comprising of Nitrogen Dioxide (NO2) and Carbon Monoxide (CO) and the visibility through the Tunnel as well. Through analysis, it is found that highest traffic recorded during the festive seasons contributed to the worst air quality inside the Tunnel but still within an acceptable rate under the Air Quality Index (AQI) Malaysia due to the use of mechanical ventilation. Artificial Neural Network (ANN) algorithm is used to train the measured data to predict fire in Meru-Menora Tunnel. If the precaution checklist is used, we can reduce the impact from worst-case scenario case studies to the Tunnel’s operation; ensuring smooth operations to the benefit of the operators and the public alike.

Abstract

Korea has selected the autonomous vehicles, smart cities, smart factories, and smart health as new growth engines by government policy based on the 4th Industrial Revolution, and organized a bureaucratic cooperation system (4th Industrial Revolution Committee, 2017.9) The...Lead moreKorea has selected the autonomous vehicles, smart cities, smart factories, and smart health as new growth engines by government policy based on the 4th Industrial Revolution, and organized a bureaucratic cooperation system (4th Industrial Revolution Committee, 2017.9) The Ministry of Land, Transport and Maritime Affairs implemented the next generation ITS pilot project (2014 ~ 17) to take the basis of technology verification, standard certification and legal system in order to prepare for nationwide expansion of C-ITS. From 2018 onwards, C-ITS demonstration projects are being sought by identifying priority services to solve traffic problems in the nation’s highways, Seoul, and Jeju, and V2X security certification demonstration for implementation of nationwide V2X wireless communication security system. The project (2019-20) is also planned. C-ITS technology developed so far includes C-ITS sub-system, message, frequency, precision positioning technology, security authentication and certification system, etc. At present, the trends in the connected car and C-ITS are classified as autonomous navigation, V2X service, and Integrated Mobility.

Abstract

Blockage of roadside bus loading area by cars and motorcycles may prevent bus entry, causing traffic jam and traffic accidents. High percentage of traffic accident fatalities are caused by cars and motorcycles following behind buses and forced to change lanes because the ...Lead moreBlockage of roadside bus loading area by cars and motorcycles may prevent bus entry, causing traffic jam and traffic accidents. High percentage of traffic accident fatalities are caused by cars and motorcycles following behind buses and forced to change lanes because the buses are not able to park in the bus loading area. Without adopting technological enforcement, such illegal behaviour is difficult to enforce because of limited police. To address the issue, Taichung City carried out a pilot project by employing advanced technologies to enforce this illegal parking behaviour in 2018. Wireless vehicle sensors with infrared and magnetic detection functions, together with license plate recognition technology, are adopted to monitor vehicles, track illegal parking, and identify particular incidents around the bus loading area. Artificial intelligent(AI) technology has been employed in this system to recognise whether the vehicle parking in the bus loading area is bus or not. If it is not, an on-site device will be activated to warn the driver and send alarms depending on lapsed time. Intelligent video analytics will automatically capture photos and video for law enforcement purposes. The results show that the system can effectively prevent drivers from illegal parking behaviour and smooth the traffic.

Abstract

Rest areas along expressways are an important traffic space for drivers. In particular, if traveling for a long period of time, there are no facilities other than rest areas where he/she can rest, and the state of development of rest facilities may be a large limiting con...Lead moreRest areas along expressways are an important traffic space for drivers. In particular, if traveling for a long period of time, there are no facilities other than rest areas where he/she can rest, and the state of development of rest facilities may be a large limiting condition in the user’s actions. Furthermore, rest facilities are equipped with toilets; they serve as not only a place where one can eliminate their waste, they also play a role in reducing physical and mental fatigue and stress caused by long periods of travel. Based on such a background, the Central Nippon Expressway Company (hereinafter “C-NEXCO ”) has developed a device that checks the state of fatigue of drivers (hereinafter “The Health Toilet”) in restrooms that are most frequented by drivers, promoting drivers to rest depending on their state of fatigue.

Abstract

Central has been the central business district of Hong Kong since the last century. It is densely-packed with high rise buildings of multi-national financial corporates, government headquarters and built heritages. Roads in the core Central are narrow with busy daily traf...Lead moreCentral has been the central business district of Hong Kong since the last century. It is densely-packed with high rise buildings of multi-national financial corporates, government headquarters and built heritages. Roads in the core Central are narrow with busy daily traffic and bustling roadside activities. The vehicular speed is getting slower and slower at daytime. To upkeep the momentum and efficiency of the road network in Central, tackling the congested traffic in the area is our top priority. Electronic road pricing (“ERP”) is a smart traffic demand management tool for alleviating local road traffic congestion. With the excellent public transport service and the commissioning of the Central – Wan Chai Bypass in January 2019, this is the right moment to take forward the ERP Pilot Scheme in Central. This paper highlights the key features of the ERP pilot scheme in Central.

Abstract

Intelligent traffic control systems have been gaining attention in an attempt to optimise traffic signal control based on the real-time traffic volume, thereby minimising delay and relieving congestion at junctions. Contrary to conventional traffic adaptive control system...Lead moreIntelligent traffic control systems have been gaining attention in an attempt to optimise traffic signal control based on the real-time traffic volume, thereby minimising delay and relieving congestion at junctions. Contrary to conventional traffic adaptive control systems where traffic signals are adjusted gradually according to the detection of the presence of vehicles approaching the junctions, intelligent traffic control systems can adjust traffic signals more instantaneously, and take pedestrian traffic into consideration. The application could be further extended to promote walkability, the use of public transport, vehicle-to-everything application, etc. Primarily, the intelligent system involves the use of sensing technology, signal optimisation algorithm and communication interface with traffic controllers. Hong Kong is a highly urbanised city and the traffic control system is an indispensable element in regulating urban traffic. This paper briefly describes the trial of adopting the intelligent traffic control system with the specific use of deep learning video analytic technique for vehicle detection at a selected location in Hong Kong.

Abstract

Supervisory Control and Data Acquisition (SCADA) systems which monitor and control physical processes/operations within a rail infrastructure are critical. SCADA systems have access to key components and infrastructure information making SCADA systems a promising attack t...Lead moreSupervisory Control and Data Acquisition (SCADA) systems which monitor and control physical processes/operations within a rail infrastructure are critical. SCADA systems have access to key components and infrastructure information making SCADA systems a promising attack target. This paper explores building machine learning models to detect anomalies in a rail SCADA system through the usage of network traffic data. The attack scenarios designed based on the domain expertise are epoch time attack and TCP payload length attack in this paper. Data pre-processing is done before passing into machine learning models for training. The anomaly detection machine learning models are evaluated using several metrics such as true positive rate and precision. Results show that supervised learning approaches (K-Nearest Neighbour, LinearSVC, Random Forest, Gaussian Bayes) outperform unsupervised learning approach (K-Means). Finally exploration into the use of the full original network versus alarm/event specific subset of network traffic for model training has shown that the latter performed better in precision due to the possible presence of overfitting to specific alarm network traffic.

Abstract

To improve the congested intersections caused by increasing heavy scooter-vehicle mixed flows in developing countries, this study presents a simulation-based signal optimization model for optimizing both the cycle length and signal timings in these intersections. The prop...Lead moreTo improve the congested intersections caused by increasing heavy scooter-vehicle mixed flows in developing countries, this study presents a simulation-based signal optimization model for optimizing both the cycle length and signal timings in these intersections. The proposed model accounts for the interactions between scooter and passenger- car flows and elaborates the maneuverability of scooters in process of parallel discharging, propagation, and formation of intersection queues. Also, the path-based signal optimization model offers an advanced algorithm to search for the optimal arterial offsets and MAXBAND, based on the complex left-turn, departure and arrival patterns of mixed traffic flows estimated with the simulation module. Our research concludes that the proposed path-based signal control model considering the unique characteristics of scooter flows can effectively conduce to congestion problems caused by heavy mixed scooter-vehicle flows.

Abstract

In China interchanges are prone to occur more crashes than basic section of freeway, especially in diverging influence area. Effective safety evaluation method for interchange is still a scarcity because of diverging and merging. Based on driving workload theory, three fi...Lead moreIn China interchanges are prone to occur more crashes than basic section of freeway, especially in diverging influence area. Effective safety evaluation method for interchange is still a scarcity because of diverging and merging. Based on driving workload theory, three field experiments were conducted on Chinese freeways, 18 professional drivers participated in tests and 102 samples were collected. Characteristics of operating speed and driving workload in interchange were analyzed, a nonlinear regression model is presented and validated. Results show that driving workload is influenced by diverging vehicles, its distribution is consistent with the distribution of crashes in diverging influence area. It is also indicated that the safety risk of mainline is mainly caused by diverging. This paper provides a new safety analysis and evaluation method for diverging area, it is a valuable reference for design optimization, traffic management and safety improvement of interchange.

Abstract

To improve the seriously congested corridors caused by significant concentrated traffic demand, this study presents advanced hierarchical regional traffic control strategies for considerably promoting the performance during morning peak hour. Also, this research accounts ...Lead moreTo improve the seriously congested corridors caused by significant concentrated traffic demand, this study presents advanced hierarchical regional traffic control strategies for considerably promoting the performance during morning peak hour. Also, this research accounts for the diverging, external metering and departing strategies on the commuting corridors with heavy accessing and merging flows. Considering the increasing traffic demand led by the enhancing discharge flow rate, the propounded tactics not only offer several alternative paths for upstream flow, but also adjust the green time and offset for optimizing MAXBAND. It can be concluded, from the field practices and statistics in this study, that the proposed strategies can effectively conduce to congestion problems resulted from increasing concentrated traffic demand on traffic corridors.

Abstract

Bus route planning, operation organization, and time dispatching can be informed by day-to-day variations in bus travel demand. In this study, the spatiotemporal origin-destination (OD) matrix is constructed using one month of smart card data from Xiamen Central City, Chi...Lead moreBus route planning, operation organization, and time dispatching can be informed by day-to-day variations in bus travel demand. In this study, the spatiotemporal origin-destination (OD) matrix is constructed using one month of smart card data from Xiamen Central City, China. A low-rank and sparse matrix decomposition method named the “GoDecomposition” (GoDec) algorithm is introduced to analyse the structure of bus travel demand. The raw OD matrix can be decomposed into three parts—a low-rank matrix, sparse matrix, and noise matrix—corresponding to three kinds of travel demand: periodic temporal flow, burst temporal flow, and random temporal flow. The low-rank matrix has periodicity; the sparse matrix appears at specific dates and locations; and the proportion of random temporal flow is slight. The reconstruction error of the low-rank matrix to the raw OD matrix is only 5.1%. The decomposition results are consistent with average daily OD matrix. However, the difference between them was large in a specific space (e.g., boarding zone/alighting zone/OD). The GoDec algorithm can be expanded to other kind of travel demand analysis, by using mobile phone data, floating car data, and so on.

Abstract

In Japan, the Ministry of Land, Infrastructure, Transport and Tourism and the expressway companies collect probe information (ETC 2.0 probe data) such as the history of the traveling position and the time etc. of each vehicle from the ETC2.0 onboard equipment. It is expec...Lead moreIn Japan, the Ministry of Land, Infrastructure, Transport and Tourism and the expressway companies collect probe information (ETC 2.0 probe data) such as the history of the traveling position and the time etc. of each vehicle from the ETC2.0 onboard equipment. It is expected that these traffic big data will be utilized to reduce traffic accidents and alleviate road congestion, and it can be utilized for efforts to reduce environmental impact. This study aimed to show the possibility of using ETC2.0 probe data for automatic operation and driving support system and described issues of using ETC2.0 probe data.

Abstract

Commercial vehicles in Korea are required to be equipped with a black box(Video Recording Device), DTG (Digital Tachograph) and ADAS (Advanced Driver Assistance System) according to law. In this study, we developed an algorithm that can process highway fog event informati...Lead moreCommercial vehicles in Korea are required to be equipped with a black box(Video Recording Device), DTG (Digital Tachograph) and ADAS (Advanced Driver Assistance System) according to law. In this study, we developed an algorithm that can process highway fog event information in real time among the information collected from probe vehicles equipped with these sensors. The algorithm includes the following functions: ①generating fog section information on an digital map, ②updating the fog section information according to the information of the following vehicles, and ③combining and deleting fog sections according to fog situation change. The algorithm can be used for the highway operator to quickly detect the position of the fog sections on the highway and to inform drivers of the start and end points of the highway fog sections to prevent traffic accidents.

Abstract

Object detection under heavy rain conditions is a challenging task that has been addressed numerous times by methods that aim to remove the ill effects caused by rain particles. While these approaches are successful in progressing the state-of-the-art in de-raining, the n...Lead moreObject detection under heavy rain conditions is a challenging task that has been addressed numerous times by methods that aim to remove the ill effects caused by rain particles. While these approaches are successful in progressing the state-of-the-art in de-raining, the notion of rainy data being directly useful in object detection is frequently over looked. In this paper we discuss the usefulness of rainy data in improving the robustness of object detectors in rainy conditions. In the process, we ascertained that detector models trained on rainy data significantly outperformed models that were trained on only clear images. We also identify that the rear lights of vehicles are powerful features that may be used to enhance the detection and tracking of vehicles. This was identified by the propagation of these features to upper layers of the network, when other distinguishing features, such as the outlines of the car were less successful in progressing up the model. Experimental results, although still slightly underperforming, were a proof of concept that the considerations in this paper do improve the overall accuracy of detection models.

Abstract

With the advent of the fourth industrial revolution era, research and development on autonomous vehicles has been actively conducted worldwide. In this international trend, many automakers and research institutes are developing SAE level 3 or higher autonomous vehicles. I...Lead moreWith the advent of the fourth industrial revolution era, research and development on autonomous vehicles has been actively conducted worldwide. In this international trend, many automakers and research institutes are developing SAE level 3 or higher autonomous vehicles. In the level 3 autonomous vehicle, takeover of the vehicle control between the driver and the vehicle is essential. This study was initiated to measure the takeover time of autonomous vehicle on a freeway. To the end, a driving simulator was used and the situation was set up similar to the freeway. A total of 50 participants participated in the experiment using the driving simulator. the takeover times were measured in various traffic volumes and curve radius during the experiment

Abstract

The automated cars only can drive safely by knowing objects existing around. The environment recognition is important for the automated driving technology. LiDAR is used as the sensor for external recognition in the recent year. The problem of LiDAR is that the farther th...Lead moreThe automated cars only can drive safely by knowing objects existing around. The environment recognition is important for the automated driving technology. LiDAR is used as the sensor for external recognition in the recent year. The problem of LiDAR is that the farther the object is, the fewer points the object can get. We solve the problem by applying stochastic resonance. Stochastic resonance is the phenomenon, which the signal detection capability improves by adding white Gaussian noise. It is aimed at generating the noise used in stochastic resonance in a point group which cannot be recognized at a long distance to reduce the recognition error of the object. The method is applied for the point cloud data at far distance. We examine the effect by applying this method. Experiment results with high rate of successful recognition with different objects prove the effectiveness of the method.

Abstract

In the self-driving system of the electric wheelchair, it is necessary to estimate the self-position of the wheelchair, identify the travelable area, and determine the traveling route. For indoor self-position estimation, location can be estimated by wireless LAN or beaco...Lead moreIn the self-driving system of the electric wheelchair, it is necessary to estimate the self-position of the wheelchair, identify the travelable area, and determine the traveling route. For indoor self-position estimation, location can be estimated by wireless LAN or beacon. However, for autonomous driving, more accurate position estimation is required. Conventionally, we have used AR (Augmented Reality) markers with high positioning accuracy as a method of self-position estimation of a wheelchair in a relatively narrow space. However, the positioning error of the angle when the AR marker was recognized from frontal direction was large, and its improvement was a problem. In this research, we propose a method to correct positioning error of markers by self-position estimation using wheelchair odometry information and object detection. Since AR marker and odometry information each have an error, we propose the method to correct the AR marker positioning by odometry information based on the error from the distance information by object detection. As a result of preliminary experiments, we showed that by using odometry for correction, positioning error is improved and wheelchair control can be done stably.

Abstract

In automated driving of level 2 and 3, driving operation may shift from system to driver (take-over). If driver’s awakening level is low at the time of take-over, this causes inadequate response to take-over and decrease in attention to surrounding environment. Therefore,...Lead moreIn automated driving of level 2 and 3, driving operation may shift from system to driver (take-over). If driver’s awakening level is low at the time of take-over, this causes inadequate response to take-over and decrease in attention to surrounding environment. Therefore, it is important to measure driver’s awakening level during automated driving. Currently, researches focusing on shallow drowsiness are insufficient. In this research, we aim to measure the driver’s shallow drowsiness. The occurrence of drowsiness makes driver’s performance worse. Then, we get driver to perform vigilance task, and analyze awakening level from the results of vigilance task. We suppose that vigilance task can detect shallow drowsiness because it is reported that awakening level affects vigilance. For evaluation of awakening level, we use facial expression evaluation and brain waves to decide which is better. In this paper, we show the validity of vigilance task to detect driver’s awakening level.

Abstract

In level 2 and 3 automated driving, take-over is occurred unexpectedly by errors of automated driving system. However driver often tends to be low awakening level during automated driving because of less work. The low arousal is one of the important tasks because there ar...Lead moreIn level 2 and 3 automated driving, take-over is occurred unexpectedly by errors of automated driving system. However driver often tends to be low awakening level during automated driving because of less work. The low arousal is one of the important tasks because there are risks that it affects the driving operation and the surrounding environment recognition. In previous research, giving the driver each task was effective for keeping awakening level in driving simulator experiment. In this research, we aim to suggest HMI to maintain awareness of driver during automated operation and use tasks that are considered to be effective for keeping awakening level in a real vehicle environment and verify task evaluation. In this experiment, mockup cockpit is used to reproduce the automatic operation of the real environment. We obtain and analyze driver’s EEG and consider effective tasks for keeping awakening level.

Abstract

Systems for recommending to electric vehicle (EV) users quick chargers along expressways are important for solving anxiety of driving range. Such systems require high-accuracy energy consumption predictions. EV energy consumption depends on factors such as the EV model, d...Lead moreSystems for recommending to electric vehicle (EV) users quick chargers along expressways are important for solving anxiety of driving range. Such systems require high-accuracy energy consumption predictions. EV energy consumption depends on factors such as the EV model, driving conditions, road type, and weather. EVs are rapidly spreading, therefore there is need for automatically constructed prediction models for energy consumption. For system safety and reliability, prediction models need to be explainable and compact to ensure understandability of prediction results. We therefore used a model tree for automatic construction of prediction models by machine learning. Here, we use actual data of over 10,000 EV trips to evaluate our prediction model. The results show that our prediction model predicts energy consumption with high accuracy in comparison with other methods and reduces tree sizes, which represent compactness of prediction model, by about 75%.