A Compressive Sensing Approach for Connected Vehicle Data Capture and Its Impact on Travel Time Estimation
Lei Lin, PARC ResearchShow Abstract
Weizi Li, University of North Carolina, Chapel Hill
Srinivas Peeta, Purdue University
Connected vehicles (CVs) can capture and transmit detailed data like vehicle position, speed and so on through vehicle-to-vehicle and vehicle-to-infrastructure communications. The wealth of CV data brings new opportunities to improve the safety, mobility and sustainability of transportation systems; however, the potential data explosion likely will over-burden storage and communication systems. To solve this issue, we design a compressive sensing (CS) approach which allows CVs to capture and compress data in real-time and later recover the original data accurately and efficiently. Two case studies are conducted to test the CS approach. For the first case study, the CS approach is applied to re-capture 10 million CV Basic Safety Message (BSM) speed samples from the Safety Pilot Model Deployment program. With a compression ratio of 0.2, it is found that the CS approach can recover the original speed data with the root mean-squared error as low as 0.05; the recovery performances of the CS approach related to other BSM variables are also explored in detail. For the second case study, a freeway traffic simulation model is built to evaluate the impact of the CS approach on travel time estimation. Multiple scenarios with various CV market penetration rates, On-board Unit (OBU) capacities, compression ratios, arrival rate patterns, and data capture rates are simulated. The simulation results show that travel times from the CS approach are more accurate for all the scenarios. It is also found that when the compression ratio of the CS approach is low, the CS approach can reach a low travel time estimation accuracy with a small CV OBU capacity. Therefore, large amounts of OBU hardware cost can be saved. Furthermore, the CS approach can greatly improve the accuracy of the travel time estimations when CVs are in traffic congestion. The reason that the CS approach has such advantages is mainly because it allows CV data to cover a longer road segment and a longer period of time, and the recovery of the original CV data is accurate and efficient.
Heuristic Methods of Inertial Signal Alignment to Enhance the Detection and Localiation Accuracy of Railtrack Anomalies
Leonard Aun Chia, North Dakota State UniversityShow Abstract
Bhavana Bhardwaj, North Dakota State University
Raj Bridgelall, North Dakota State University
Pan Lu, North Dakota State University
Denver Tolliver, Upper Great Plains Transportation Institute
Neeraj Dhingra, North Dakota State University
Maintaining the designed geometry of railroad tracks is vitally important for the smooth and safe passage of vehicles. Uneven track surface can result in poor ride quality and possible derailments. Railroad companies currently cannot afford to monitor the geometry of their entire networks as frequently as necessary because of the high cost of deploying the required specialized equipment and trained professionals. The use of low-cost sensors aboard railcars could screen the infrastructure for anomalies automatically and continuously to save railroad companies billions of dollars by focusing follow-up manual inspections on high-risk locations. Smartphones now have all of the sensor capabilities needed to test and validate a low-cost condition monitoring system. On-board sensors allow for the combining of features extracted from inertial sensor signals, across multiple train traversals, to significantly enhance their signal-to-noise ratio, which minimizes false positives and false negatives. However, position registration errors, relatively slow update rates of the GPS receivers on low-cost devices, and non-uniform sample interval variabilities of their inertial sensors result in feature alignment errors that can actually degrade the signal-to-noise ratio. This paper introduces four heuristic methods to align the inertial signals from multiple traversals. The authors selected the best-performing method by ranking the position variability of a known ground truth anomaly and the corresponding statistical distribution that maximally agrees with a Gaussian.
Points Registration for Roadside LiDAR Sensors
Jianqing Wu, University of Nevada, RenoShow Abstract
Hao Xu, University of Nevada, Reno
Wei Liu, Chongqing Jiaotong University
The roadside LiDAR deployment provides a solution to obtain the real-time high-resolution micro traffic data (HRMTD) of unconnected road users for the connected-vehicle road network. Single roadside LiDAR sensor has a lot of limitations considering the scant coverage and the difficulty of handling object occlusion issue. Multiple roadside LiDAR sensors can provide a larger coverage and eliminate the object occlusion issue. To combine different LiDAR sensors, it is necessary to integrate the point clouds into the same coordinate system. The existing points registration methods serving mapping scans or autonomous sensing systems could not be directly used for roadside LiDAR sensors considering the different feature of point clouds and the spare points in the cost-effective roadside LiDAR sensors. This paper developed an approach for roadside LiDAR points registration. The developed points-aggregation-based partial iterative closest point algorithm (PA-PICP) is a semi-automatic points registration method, which contains two major parts: XY data registration and Z adjustment. A semi-automatic key point selection method was introduced. The partial iterative closest point (PICP) was applied to minimize the difference between different LiDARs in the XY plane. The intersection of ground surface between different LiDARs was used for Z-axis adjustment. The performance of the developed procedure was evaluated with field collected LiDAR data. The results showed the effectiveness and accuracy of data integration using PA-PICP was greatly improved compared to points registration using the traditional ICP. The case studies also showed that the occlusion issue can be fixed after PA-PICP points registration.
A Framework for Network-Level Pavement Condition Assessment Using Remote Sensing Data Mining
Stefanos Politis, University of Texas, AustinShow Abstract
Zhanmin Zhang, University of Texas, Austin
Sareh Kouchaki, University of Texas, Austin
Carlos Caldas, University of Texas, Austin
Pavement condition monitoring is essential for efficient resource allocation in transportation asset management. However, the collection of data involves laborious and costly procedures. The intention of this study is to investigate the usage of remote sensing data for network level pavement condition assessment in order to provide a more cost-effective alternative. For this reason, an extensive literature review has been conducted and a data mining framework has been established utilizing the inherent information of multispectral orthoimages in order to train models that will be capable of predicting the pavement condition of different road segments. A preliminary case study was conducted with data provided by the City of Dallas and remote sensing images acquired from the Texas Natural Resources Information System. An image segmentation algorithm was employed to separate pavements from other surfaces, and three different classifiers were compared for pavement condition class prediction. The results showed that up to 67 percent of the sampled segments were correctly classified, indicating that the framework might have potential for future implementation if further research is conducted on the different constituent steps in order to increase classification accuracy.
Automated Data Collection and Safety Analysis at Intersections Based on a Novel Video Processing System
Qiuchen Zhang, Tongji UniversityShow Abstract
Zhen Yang, Tongji University
Dazhi Sun, Texas A&M University, Kingsville
A novel video-based system is presented which collects trajectories and motion parameters of all objects at intersections. First, a modified ViBe method is used to extract the foreground of moving objects. Then, an Object-Point-Contour (OPC) matching approach is developed for pairing, tracking and generating trajectories. Finally, raw trajectories are corrected through post-processing and motion parameters are estimated after object classification. This system demonstrates better performance while tracking tardy and shadowed objects. compared with previous studies. The accuracy of 86% and 91% are obtained for traffic counts and velocity validation, respectively. This paper also presents a sample safety analysis using Traffic Conflict Technology (TCT) to demonstrate the possible implementation for traffic management and safety analysis.
A Proactive Approach for Intersection Safety Visualization Based on Real-Time Radar Sensor Data
Muting Ma, University of LouisvilleShow Abstract
Zhixia Li, University of Louisville
Intersection crashes constitute a significant portion of total crashes nationwide, which amount to about 44 percent of all reported crashes. Traditional methods of this type of crash visualization include intersection collision diagram and GIS-based intersection crash hot spot map. Since all these methods are based on historical crash data, such collision diagrams and intersection crash hot spot maps will only be created until crash happens. It is necessary to convert these reactive method to be proactive before real crashes happen. An approach that can identify, classify, quantify, and cluster traffic conflicts with various types at intersection using real-time radar sensor data was developed. Using this approach, the traffic conflicts data that will be visualized in the Intersection Proactive Safety Visualization system are actually true representation of the intersection safety issues, rather than commonly viewed as surrogate safety measures. The results of a pilot study have shown the potential and verified the feasibility of developing IPSV with visulation of traffic conflicts of different conflict types as well as visulation of contour/head map showing conflict severity’s locational distribution. The system is unique as it identifies traffic safety issues by using a proactive safety measure reflected by traffic conflicts and its severity assessed by Time-to-Collision.
Video Analytics for Estimating Control Delays at Signalized Intersections Based on Videos Collected by Unmanned Aerial Vehicles
Pengmin Pan, Auburn UniversityShow Abstract
Chennan Xue, Auburn University
Huaguo Zhou, Auburn University
Control delays of signalized intersections are often estimated based on Highway Capacity Manual (HCM) methods or using traffic simulation tools. Both methods require three inputs – turning movement counts, geometric features, and signal timing parameters. Collecting these data can be a time-consuming and labor-intensive task. The application of Unmanned Aerial Vehicle (UAV) has gained particular attention in the transportation infrastructure industry. On a nationwide scale, there has been a trend to incorporate UAV in traffic monitoring and operational analysis. However, UAV has not yet been applied directly in the performance measurement at signalized intersections. In this study, a UAV video analytic method was developed to determine control delays at signalized intersections. The core algorithm is to apply background subtraction to detect vehicles and frame overlapping to determine stopped ones. By recording the frame number and position of each stopped vehicle found, the control delay is calculated as the time difference between vehicle stop and reaccelerating. Comparing the calculated average lane delay with simulation outputs (Synchro software), the maximum difference is 8.9%. The result indicates that the developed video analysis algorithm can provide precise estimates of control delays at signalized intersections as an alternative cost-effective approach. Moreover, extra information such as distributions of control delays in each queue can be determined, which improves the accuracy of the intersection analysis.
An Automatic Ground Points Identification Method for Roadside LiDAR Data
Jianqing Wu, University of Nevada, RenoShow Abstract
Hao Xu, University of Nevada, Reno
Bin Lv, Lanzhou Jiaotong University
Rui Yue, University of Nevada, Reno
Yang Li, Beijing Institute of Petrochemical Technology
The roadside Light Detection and Ranging (LiDAR) provides a solution to fill the data gap under the mixed traffic situation. The real-time high-resolution micro traffic data (HRMTD) of all road users from the roadside LiDAR sensor provides the new opportunity to serve the connected-vehicle system during the transition period from unconnected vehicles to connected vehicles. The ground surface identification is the basic data processing step for the HRMTD collection. The current ground points identification algorithms based on airborne and mobile LiDAR did not work for roadside LiDAR. A novel algorithm is developed in this paper to identify and exclude ground points based on the features of LiDAR, terrain and point density in the space. The scan feature of different beams is used to search ground points. The whole procedure can be divided into four major parts: points clustering in each beam, slope-based filtering, shape-based filtering, and ground points matrix extraction. The proposed algorithm was evaluated using the real-world LiDAR data collected at different scenarios. The results showed that this algorithm can be used for ground points exclusion under different situations (differing terrain types, weather situations, and traffic volumes) with high accuracy. This algorithm was compared with the previously developed algorithms. The overall performance of the proposed algorithm is superior. The low computational load guarantees this method may be applied for the real-time data processing.
Toward Compressed Indexing of OpenStreetMap Data
Pengfei Yi, Beihang UniversityShow Abstract
Sebastian Wandelt, Beihang University
Xiaoqian Sun, Beihang University
Research studies on accessibility, urbanization, and land usage, increasingly make use of the community-driven project OpenStreetMap(OSM). The amount of data collected throughout the world via OSM has grown tremendously throughout the last decade, with the current planet-scale dataset reaching almost one terabyte of uncompressed storage (in XML format). Except from the standard XML format, OSM comes, essentially, in two flavors: compressed (PBF) and indexed (Overpass). The former reduces the data to around 35 GB, but does not support efficient query answering, given the lack of random access and indexing support. The latter, on the other hand, provides fast query answering at the price of high storage costs for index data structures. For instance, the OSM subset for China alone requires around 360 MB storage in PBF, but comes at a staggering 17 GB for Overpass. In this paper, we propose a novel compressed index structure for managing OSM data. The main ideas for compression are spatial partitioning and a column-wise storage strategy. We provide efficient meta-data access by a variant of bloom filters, tailored towards our column-based block storage infrastructure. Our novel index structure is evaluated on several large-scale datasets, including Berlin, South Korea, China, US-Northeast and US-Midwest. Our experimental results indicate that our compression-based indexing technique provides a nice sweet spot, balancing fast query answering with small storage costs.
The Use of Aerial LiDAR in Measuring Streetscape and Street Trees
Yaneev Golombek, University of Colorado, DenverShow Abstract
Wesley Marshall, University of Colorado, Denver
This paper investigates the usefulness of 3D volumetric pixels (voxels) and USGS Quality Level 2 (QL2) LiDAR data to measure features in streetscapes. As the USGS embarks on a national LiDAR database with the goal of covering the entire US with QL2 data or better, this paper investigates uses of QL2 LiDAR for 3D measuring of streetscapes. Tree mapping is a common use of QL2 LiDAR data, and street trees are among the most common features within urban streetscapes that transportation and urban designers assess. Traditional remote sensing techniques derive tree polygons from imagery, and traditional uses of LiDAR for tree canopy mapping is based on deriving a 2D canopy polygon with an attribute for elevation height. However, when breaking up streetscapes into 5 ft elevation zones and calculating street-tree voxels at each elevation zone height, 3D characteristics of street trees become prevalent that completely differ from the common 2D LiDAR derived street trees. Statistical tests in this paper display how different the 3D characteristics are from the 2D derived LiDAR Polygons, as this paper introduces a new methodology for measuring streetscape features in 3D, particularly street trees.
Developing an Aerial Image–Based Approach for Sidewalk Digitalization
Ji Luo, University of California, RiversideShow Abstract
Guoyuan Wu, University of California, Riverside
Zhensong Wei, University of California, Riverside
Kanok Boriboonsomsin, University of California, Riverside
Matthew Barth, University of California, Riverside
To support the active mobility, extensive work has been focused on planning, maintaining and enhancing the infrastructure, such as sidewalks. A significant amount of these efforts have to go for the setup and maintenance of the sidewalk inventory on a certain geographic scale (e.g., citywide, statewide). To address the stated problem, it is proposed herein to develop an aerial-image-based approach that can: 1) extract the features of sidewalks based on digital vehicle road network; 2) overlay the initial sidewalk features with aerial imagery and extract aerial images around sidewalk area; 3) apply a machine learning algorithm to classify sidewalks images into two major categories, i.e., concrete-surface present or sidewalks missing and 4) construct a connected sidewalk network in a time-efficient and cost-effective manner. A deep convolutional neural network is applied to classify the extracted sidewalk images. The learning algorithm achieves 97.22% total predication rate for the test set and 92.6% total predication rate in the blind test. The proposed method takes full advantage of available data sources and build on top of existing roadway network to digitize sidewalk.
The Use of Multi-Rotor Unmanned Aerial Vehicles for Fine-Grained, Roadside Air Quality Monitoring
Bai Li, Shanghai Jiao Tong UniversityShow Abstract
Rong Cao, Shanghai Jiao Tong University
Zhanyong Wang, Fujian Agriculture and Forestry University
Rui-Feng Song, Shanghai Jiao Tong University
Zhong-Ren Peng, Shanghai Jiao Tong University
Guangli Xiu, East China University of Science and Technology
Qingyan Fu, Shanghai Environmental Monitoring Center
With the increasing number of motor vehicles, vehicle exhaust gas has gradually become one of the important sources of urban air pollutants. After being exhausted from the motor vehicle, the exhaust gas spreads along the air on the road and gradually deposits into the surrounding area, which has an adverse impact on pedestrians and residents. At present, most researches on vehicle exhaust are directly measuring the total emissions from exhaust pipe or monitoring the time variation of air pollutants in the roadside by setting roadside monitoring stations. However, the spatial resolution of these two methods is very low, and it is impossible to accurately describe the diffusion pattern of exhaust gas in the atmosphere after discharge. In recent years, some scholars have conducted research on the quality of roadside air by hand-held portable devices, but these are limited by the speed of people's travel, and the spatial and temporal resolution of the acquired data is also very low. By using multi-rotor unmanned aerial vehicles (UAVs) and portable equipment, this study demonstrates the atmospheric environment monitoring system based on multi-rotor UAV. Benefiting from the flexible requirements of takeoff or landing sites and high maneuverability of multi-rotor UAVs, the system has increased the capability for higher resolution spatial and temporal monitoring of traffic emitted pollutants diffusion. The system can carry out fixed-point measurement by hovering, and can also measure air pollutants in complex urban terrain, which provides a new idea for the study of vehicle exhaust gas diffusion mode.
An Advanced Framework for Traffic Parameters Estimation from UAV Video
Ruimin Ke, University of WashingtonShow Abstract
Shuo Feng, Tsinghua University
Zhiyong Cui, University of Washington
Yinhai Wang, University of Washington
Currently unmanned aerial vehicle (UAV) is at the heart of traffic sensing research due to its advantages such as low cost, high flexibility, and wide view range over traditional traffic sensing technologies. It opens up new opportunities for intelligent transportation systems by supporting efficient and reliable traffic monitoring and decision making. Under such context, an increasing trend of emerging research on UAV-based traffic detection can be observed. Recent studies have already achieved great progress on the extraction of aggregated macroscopic traffic parameters from UAV videos, however, there is still a wide gap between the state-of-the-art methods and a complete framework that can automatically estimate both microscopic traffic parameters and lane-level macroscopic traffic parameters at the same time. In this paper, an advanced framework is proposed to fill this gap by addressing several key challenges. This framework is composed of three functional modules: core functional module, data storing module, and traffic parameters estimation module. The core functional module consists of two sub-modules, which are a new method for lane detection and an exclusive method for multiple vehicle tracking in UAV video. The integration of the two sub-modules enables the collection of various raw traffic data, which are then organized and stored in the data storing module for further processing. The traffic parameters estimation module is designed to calculate the macroscopic and microscopic traffic parameters, and to analyze individual vehicle behaviors. Experiments on real-world UAV video data and thorough analyses on the results demonstrate the promising performances of the proposed framework.
Microscopic Traffic Parameter Extraction from Aerial Videos with Multi-Dimensional Camera Movements
Zhibin Li, Southeast UniversityShow Abstract
Xinqiang Chen, Shanghai Maritime University
Lei Ling, China Design Group Co., Ltd
Huafeng Wu, Shanghai Maritime University
Wenzhu Zhou, Southeast University
Unmanned Aerial Vehicle (UAV) is becoming increasingly popular in traffic monitoring due to its low cost, wide view coverage, and rapid deployment. Extracting traffic parameters from UAV videos can provide important information which benefits traffic flow analysis and control. Most of previous studies focused on extracting aggregated traffic parameter planar movement of UAV cameras. Little attention has been paid on the videos recorded with multi-dimensional camera movement. To address this issue, this paper proposes a novel framework for traffic parameter extraction with UAV camera moving at different dimensions. We first employ a temporally robust global motion compensation (TRGMC) model to compensate UAV camera movements and obtain stabilized background. Then, the kernelized correlation filter (KCF) is applied to track vehicles fast and accurately. After that, we employ the Hough line detection to find out reference markings and map the image length to physical length. Finally, microscopic traffic parameters including individual vehicle speed, time headway and space headway in a traffic stream are estimated using outputs from previous steps. The experimental results show that the proposed method achieves an accuracy of about 94.60% and 93.94% in estimating vehicle speed and headway, respectively.
Change Point Models for Real-Time V2I Cyber Attack Detection in a Connected Vehicle Environment
Gurcan Comert, Benedict CollegeShow Abstract
Mizanur Rahman, Clemson University
Mhafuzul Islam, Clemson University
Mashrur Chowdhury, Clemson University
Connected vehicle (CV) systems are cognizant of potential cyber attacks because of increasing connectivity between its different components such as vehicles, roadside infrastructure and traffic management centers. However, it is a challenge to detect security threats in real-time and develop appropriate/effective countermeasures for a CV system because of the dynamic behavior of such attacks, high computational power requirement and a historical data requirement for training detection models. To address these challenges, statistical models, especially change point models, have potentials for real-time anomaly detections. Thus, the objective of this study is to investigate the efficacy of two change point models, Expectation Maximization (EM) and Cumulative Sum (CUSUM), for real-time V2I cyber attack detection in a CV Environment. To prove the efficacy of these models, we evaluated these two models for three different type of cyber attack, denial of service (DOS), impersonation, and false information, using basic safety messages (BSMs) generated from CVs through simulation. Results from numerical analysis revealed that EM and CUSUM could detect these cyber attacks, DOS, impersonation, and false information, with an accuracy of 99%, 100%, and 98%, and 100%, 100% and 98%, respectively.
A Novel Framework for Automatic Vehicle Trajectory Extraction and Denoising from Aerial Videos
Xinqiang Chen, Shanghai Maritime UniversityShow Abstract
Zhibin Li, Southeast University
Yongsheng Yang, Shanghai Maritime University
Huafeng Wu, Shanghai Maritime University
Ruimin Ke, University of Washington
Wenzhu Zhou, Southeast University
In recent years, unmanned aerial vehicle (UAV) has become an increasingly popular tool for traffic monitoring and data collection on highways due to its low cost, high resolution, good flexibility, and large coverage. Extracting high-resolution vehicle trajectory data, which provides wide support for both microscopic and macroscopic traffic flow analysis, from aerial videos taken by UAV flying over road section becomes a critical research task. In this study, we propose a novel methodological framework for automatic and accurate vehicle trajectory and length extraction from aerial videos. We first employ an ensemble detector to detect vehicles on the target region. Then the kernelized correlation filter (KCF) is applied to track vehicles in a fast and accurate way. The vehicle positions are mapped from the physical coordinates to the Frenet coordinate to obtain the vehicle trajectories along the road. The data quality control is applied in the procedure and a Wavelet Transform is used to denoise the biased vehicle positions in the trajectory data. Our model was tested on two aerial videos on freeway segments. The experimental results show that the proposed method extracts vehicle trajectory at a high accuracy (i.e., measurement error of Mean Squared Deviation (MSD) is 28.854 pixels, Root-mean-square deviation (RMSE) is 2.187 pixels, the Pearson product-moment correlation coefficient (Pearson's r) is 0.999), which provides us a reliable trajectory for analyzing traffic flow. This study fills gaps in UAV-based automatic vehicle trajectory extraction, and has the potential to benefit a variety of future research.
Expanding the Capabilities of Radar-Based Vehicle Detection Systems: Noise Characterization and Removal Procedures
Kelvin Santiago, University of Wisconsin, MadisonShow Abstract
David Noyce, University of Wisconsin, Madison
The capabilities of radar-based vehicle detection (RVD) systems used at signalized intersections for stop bar and advanced detection are arguably underutilized. Underutilization happens because RVD systems can monitor the position and speed (i.e. trajectory) of multiple vehicles at the same time but these trajectories are only used to emulate the behavior of legacy detection systems such as inductive loop detectors. When full vehicle trajectories tracked by an RVD system are collected, detailed traffic operations and safety performance measures can be calculated for signalized intersections. Unfortunately, trajectory datasets obtained from RVD systems often contain significant noise that makes the computation of performance measures difficult. In this paper, a description of the type of trajectory datasets that can be obtained from RVD systems is presented along with a characterization of the noise expected in these datasets. Guidance on the noise removal procedures that can be applied to these datasets is also presented. This guidance can be used by those that want to use data from commercially-available RVD systems to obtain advanced performance measures. To demonstrate the potential accuracy of the noise removal procedures, the procedures were applied to trajectory data obtained from an existing intersection, and a basic performance measure (vehicle volume) data were extracted from the dataset. Volume data derived from the de-noised trajectory dataset was compared with ground truth volume and an absolute average difference of approximately 1 vehicle every 5 minutes was found, thus highlighting the potential accuracy of the noise removal procedures introduced.
Blockchain: A Safe, Efficient Solution for Driver Privacy and Connected Vehicle Transportation Data Sharing
Yingxi Cao, New York UniversityShow Abstract
Abdullah Kurkcu, New York University
Kaan Ozbay, New York University
Connected and automated vehicles (CAVs) are becoming increasingly prevalent, bringing with them potential for better safety and mobility. However, these vehicles can create many thousands of transactions in a flash, creating a challenge for current technologies that are not capable of transmitting such big data “privately” and “securely”. Distributed ledger technologies such as Blockchain have the potential to address this challenge by using decentralized system. Blockchain-based system allows users to enter into direct relationships with each other following commonly agreed terms with a high degree of trust, eliminating the need for a central authority while retaining security and privacy. This study investigates the potential for Blockchain to support safer delivery of CAV data. By using an actual simulation based implementation of the proposed architecture, it demonstrates how Blockchain can improve the level of security and privacy of data sharing and attempts to answer two fundamental questions: 1) how to securely and privately get and store data from CAVs and 2) how to find the best method to connect them using Blockchain technology. The proposed eight-layer framework uses Hyperledger Fabric as an underlying Blockchain technology and uses machine learning models for analyzing data collected in chain. The traffic data in the physical layer are simulated using microscopic traffic simulation tool SUMO and then incorporated into the Blockchain platform. The experiments highlight that the CAV system can be effectively combined with Blockchain technologies while enhancing security in a significant manner.
A Novel Methodology to Derive Vehicle Occupancy Using Wi-Fi Sensors Under Heterogeneous Traffic Conditions
Ninad Gore, Sardar Vallabhbhai National Institute of Technology, SuratShow Abstract
Shriniwas Arkatkar, Sardar Vallabhbhai National Institute of Technology, Surat
Joshi Gaurang, Sardar Vallabhbhai National Institute of Technology, Surat
Ashish Bhaskar, Queensland University of Technology
Possibility of multiple patches from a single vehicle or vehicle occupancy subsides the effectiveness of Wi-Fi media access control (MAC) passive data collection technologies in real-time estimation of traffic-state. Therefore, it becomes imperative to derive vehicle-occupancy values to arrive at unbiased estimation of traffic-state. To comprehend this objective, data was collected using Wi-Fi based sensors for two different roadway and traffic characteristics. Videography was simultaneously conducted for the subject survey locations to apprehend the performance of the adopted ITS technology. Investigation revealed a wide-variation in sensor-recorded volume for a given actual traffic volume, which may be jointly accredited to higher sensor efficiency and vehicle occupancy. To derive occupancy values, primarily, a standard detection was assumed for all traffic volume levels, which was then varied to account for variation in recorded volume. Thereafter, based on standard detection, detection ratio-matrix was obtained. To arrive at occupancy value for a given traffic volume, 95th percentile of the detection ratio was obtained. The derived occupancy values, when plotted against traffic volume, revealed wide variation in occupancy values for the same traffic volume levels. Face-to-Face interview, using questionnaires to measure occupancy values, was performed and revealed that the occupancy values obtained through questionnaire and empirically developed methodology were found to be consistent. This corroborates the robustness of the developed methodology, especially under heterogeneous traffic conditions. As an important outcome of the study, occupancy-prediction model is developed, which exhibits a promise for planners and engineers to obtain occupancy values, at a given Wi-Fi penetration rate.
Automating Traffic Video Analysis for Intersection Safety Device Programs: Two Case Studies from Canadian Cities
Mohamed Zaki, University of Central FloridaShow Abstract
Tarek Sayed, University of British Columbia
Shewkar Ibrahim, City of Edmonton
This paper demonstrates the application of automated video analysis for Intersection Safety Device (ISD) programs. Computer vision (CV) is a versatile tool for traffic analysis. With the ability of accurate speed measurements and tracking vehicle coordinates, red-light violations and speed enforcements are practically possible using CV. Video analysis is a non-invasive tool that can also be used to identify and analyze traffic components like phases of traffic lights, road geometry, and lane positions. Two studies using CV are presented in the paper. The first study considers automated video analysis as an evaluation tool for three deployed ISDs in the City of Edmonton, Alberta. The evaluation is performed by comparing the speed, and enforcement data from the ISD log files and the corresponding CV tracking output. The second study considers the automated video analysis as guidance for selecting potential locations for the deployment of ISDs. Red-light and Speed violation automated detection is applied to video data collected from two intersections in the City of Fredericton, New Brunswick. Validations against manual observations are also provided to demonstrate the accuracy of the CV technology based on three factors: 1) Capability for accurate measurement of crossing speed 2) Ability to detect all red-light violations 3) Ability to detect all speeding violations.
A Rubric-Driven Evaluation of Open Data Portals and Their Data in Transportation
Archana Venkatachalapathy, Iowa State UniversityShow Abstract
Anuj Sharma, Iowa State University
Skylar Knickerbocker, Iowa State University
Neal Hawkins, Iowa State University
In recent years, the open data movement is gaining momentum in the transportation industry with multiple State's Department of Transportation (DOT) launching their own repository of datasets. The quality of data, ease of usage and availability of metadata varies from source to source. There is an imminent need to assess the quality of open data portals to provide agencies a yardstick to measure their performance and draw inspirations from higher ranking portals. This paper proposes a data portal evaluation rubric (DPER) which can serve this purpose. DPER is designed to capture the essence of the National Open Data Policy. The paper then uses DPER to evaluate 40 data portals at the state and national level which provide transportation datasets. DPER evaluates the quality of the portal, the openness of data, and the relevance of its content to the transportation sector. The portal of the State of Iowa scores the highest due to its user-friendly interface with interactive visualization tools, relevant data content and useful API references for application developers.
Imputation of Missing Transfer Passenger Flow with Self-Measuring, Multi-Task Gaussian Process
Wenhua Jiang, Monash UniversityShow Abstract
Nan Zheng, Monash University
Paul Reichl, Institute of Railway Technology
Inhi Kim, Monash University
Transportation data is of great importance for intelligent transportation control and management. Since current big data technologies can provide vast amounts information on the status of traffic systems, data collection may be interrupted by technical failures and other practical issues. There is however, an increasing demand to carry out traffic analysis when a significant amount of data is missing. This study introduces the use of a self-measuring multi-task Gaussian process (SM-MTGP) method for imputing missing data. Particularly, the study focuses on the transfer passenger flow at a railway station collected by WiFi sensors. SM-MTGP exploits the temporal correlation between observations. First, correlations between tasks and inputs are learned simultaneously with the constructed two-way array. Second, covariances of features of these two aspects are measured nonlinearly with selected kernel functions. Third, additional knowledge is provided by the responses, helping get more insights into the similarities of the observations. The performance of the proposed method is assessed under several different missing data scenarios for a large-demand railway station in Melbourne using 6-months of real data. Test results show that: (i) the SM-MTGP has the lowest imputation errors; (ii) the SM-MTGP presents substantial improvement in reducing RMSE values by 60% over the base model; (iii) the SM-MTGP in particular outperforms the conventional approaches with large missing ratios. It is motivating to find the proposed method can achieve improved performance, our on-going effort is given on incorporating other features into this algorithm that can make further application on large-scale transit network analysis.
A New Method for Traffic Flow Data Collection Using Onboard Monocular Camera
Yifan Zhuang, University of WashingtonShow Abstract
Ruimin Ke, University of Washington
Yinhai Wang, University of Washington
Traffic data collection is the fundamental step in most applications of intelligent transportation systems (ITS). Recently, with the fast development of traffic sensing technologies, traffic data collection methods have become more and more robust and diversified, yet still have some limitations in their flexibility and coverage. Onboard monocular camera, which is normally used as driver recorder, has a considerable potential to be turned into a cost-effective moving traffic sensor with its low cost and ego-vehicle’s high mobility. Existing studies have explored the feasibility of onboard camera on scene understanding, safety surrogate data collection, etc., however, few studies have been conducted to utilize onboard monocular camera for traffic flow data collection. To this end, this paper puts forward a method using the onboard monocular camera to collect traffic data. The basic structure is composed of a You-Only-Look-Once (YOLO) model and Spatial Transformer Network (STN) to detect vehicles in real-time. Then the traffic flow parameters are computed via fundamental optic theories and traffic flow definitions. The experiment results display a similar sensing accuracy to inductive loop detectors. In addition, the STN-YOLO model has a better vehicle detection rate than the original YOLO model which is pre-trained on the COCO dataset.
Research on Traffic Recognition Method Based on Noise Spectrum View Analysis
Qinglu Ma, Chongqing Jiaotong UniversityShow Abstract
Zheng Zou, Chongqing Jiaotong University
Saleem ULLAH, Khwaja Fareed University of Engineering & Information Technology
Noise signals analysis is one of effective approach in collecting traffic data, while it is hard to work when the vehicle-signal-segments exist overlaps. Conventionally, the Short Time Energy (STE) and the Short Time Average Mmagnitude (STAM) method really are helpless choice in processing of traffic detection due to both of them cannot accurately distinguish overlapped signals. To improve the detection accuracy of traffic, an innovative method, called Spectrum View (SV), is proposed to deal with the overlapped signals. Firstly, vehicle noise signals that contain the overlapped vehicle-signal-segments are collected on a single lane, and the Least Mean Square (LMS) adaptive filter is constructed to reduce noises. Next, the STE and STAM features of vehicle noise signals are extracted and smoothed. Finally, based on the STE, STAM and SV methods respectively, the endpoints of the vehicle-signal-segments are detected and the results of traffic volume monitoring are compared under different methods. The experimental results indicate: If overlap does not exist between two adjacent vehicle-signal-segments, the traffic volume monitoring results that dectected under the STE or STAM method are the same as the SV method; on the contrary, compared with the STE or STAM method, the overlapped vehicle-signal-segments can be correctly detected based on the SV method, and the accuracy of detection results is increased by 28% under the SV method.
Detecting the Roadworks from OEM Sensor Data: An Application for Autonomous Vehicle
Leon Stenneth, HERE TechnologiesShow Abstract
Xuemei Chen, Beijing Institute of Technologies
Zhenhua Zhang, HERE Technologies
This paper presents a pipeline of use the vehicle sensor data to detect the roadworks. The roadwork detection from vehicle sensor can be real time and alert the autonomous vehicle to convert to manual operation ahead of roadworks. The vehicle sensor data employed in this study is the yellow lane marking data, which is an exclusive indicator of roadwork in some Western Europe countries. Our proposed methods aim at remove the outliers, reconstruct the routes of the roadworks and pinpoint the start and end of roadworks. By comparing with the validation drive in 37 roadwork sites in Western Europe, our proposed methods can cover over 95% of the roadwork extensions on freeway and over 75% of roadwork extensions on non-freeway. The average start and end offset errors are 7.2% and 8.1% on freeways, and 9.7% and 21.3% on non-freeway. Our study unveils some key features of the vehicle sensor data and also provide insights for the roadwork detection studies from floating sensors.
Automated Pavement Friction Estimation Using Mobil LiDAR
Yishun Li, Tongji UniversityShow Abstract
Yuchuan Du, Tongji University
Yu Shen, Tongji University
Rapid detection and maintenance of pavement friction is essential for roadway crashes prevention. Traditional measurement methods are time-consuming, labor-intensive and inefficient. This paper proposes a new method to rapidly estimate pavement friction by using Light Detection and Ranging (LiDAR) sensor. Eight parameters are developed to capture the texture and material information of pavement. British Pendulum Number (BPN) is adopted as the reference of pavement friction with the data collection and processing approaches stated. An ordered logit regression model is utilized to estimate the level of pavement friction, with an average accuracy of 75.86%. The model shows that both textures and material information contribute to pavement friction. Some experimental tests are conducted to explore the potential impact of illumination, showing that lighting and road shading do not affect measurements. The proposed LiDAR-based method is able to assist for rapid, economical, and automatic estimate pavement friction.
Detection Range Analysis of Roadside LiDAR Sensors
Junxuan Zhao, Texas Tech UniversityShow Abstract
Hao Xu, University of Nevada, Reno
Jianqing Wu, University of Nevada, Reno
Ciyun Lin, Jilin University
Hongchao Liu, University of Bologna
Light Detection and Ranging (LiDAR) is a new three-dimensional technology that offers a revolutionary approach to collecting data in transportation. Currently, the on-board LiDAR sensing system in autonomous/connected vehicles is the main application of LiDAR sensors in intelligent transportation systems. Curbside infrastructure-based LiDAR systems open a new field for the research and manufacturing of LiDAR sensors. Unlike the on-board LiDAR sensors, roadside LiDAR sensors must be installed at a fixed location in order to cover an optimal range of road users. This paper presents a detection range analysis of roadside 360-degree LiDAR sensors based on the sensor’s features and deployment. Configuration and deployment recommendations on collecting pedestrian and vehicle data using roadside LiDAR sensors with and without considering the occlusion issue are proposed. The suggested installation and specification of roadside LiDAR sensors provide an important guidance and reference for both sensor users and manufacturers.
Roadside Sensing Information Enabled Horizontal Curve Crash Avoidance System
Song Wang, University of LouisvilleShow Abstract
Zhixia Li, University of Louisville
Horizontal curves are a major cause of road departure crashes that lead to fatal and severe injuries. Existing curve crash avoidance systems are typically enabled by displaying safety messages via in vehicle heads-up/down display. However, these systems just pass the information to drivers. The final decision to reduce or maintain the speed in response to safety message is still subject to human drivers. Due to the involvement of human factors, there is potential that road departure crashes will happen if human drivers do not respond to safety messages appropriately. The autonomous vehicle technology targets at eliminating human errors in driving through automated driving system. In this context, this paper proposes a conceptual prototype of a connected and autonomous vehicle-based horizontal curve crash avoidance system (CAV-HCCAS), aiming at achieving a permanent solution to the horizontal curve safety by excluding human errors through automated driving. In CAV-HCCAS, a roadside sensor detects pavement wetness level at the horizontal curve and communicates to the vehicle via Dedicated Short-Range Communication (DSRC). By processing the pavement wetness information, the autonomous vehicle applies a safe curve travel speed that reflects the real-time pavement conditions. An automated driving simulation experiment was performed to prove the concept. Dry and wet pavement conditions were simulated for a horizontal curve. Lane deviation data and the resulting lane departure conflicts were measured as safety performance measures. Results indicate a significant reduction of lane departure conflicts when CAV-HCCAS is implemented under both dry and wet pavement conditions, which reflects a substantial safety benefit.
Law of Sensor Errors Propagation in Sensor Location Flow-Observability Problem Based on Turning Ratios at Intersections
Minhua Shao, Tongji UniversityShow Abstract
Chenyang Zhou, Tongji University
Sensor location flow-observability problem is to identify the minimum set of links to be installed with sensors that allow the full inference of flows on all links. Inevitably, the measurement errors of observed link flows will accumulate and propagate on the network and affect the result of link flow inference. Investigating the law of error propagation is critical for error control in the link flow inference and optimization of sensor location. This paper analyzes the law of sensor errors propagation by using turning ratios at intersection as priori information. Firstly, the error effect is defined to reflect the influence of single sensor error on the result of flow inference of the target link. Then proving that, for the scheme installing all sensors on only entrance links, the error effect of single sensor error is the sum of product of turning ratios of all paths between the inferred link and the error source link and its error source coefficient. Besides, the scheme installing all sensors on only entrance links is equivalent to any sensor location scheme, and their sensor errors have the same effect on all links of the network. Finally, a simple grid road network and an actual road network are selected in the case study. The law of error propagation and equivalent substitution is verified. Furthermore, we find the advantages of the scheme installing all sensors on only entrance links and the theory of equivalent substitution can be applied to the optimization of sensor location in practice.