Session to cover the top reviewed papers for presentation submitted to the Committee on Geographic Information Science and Applications (ABJ60).
An Improved Automated Method for Roadway Horizontal Curve Identification Using GIS Data
Ilir Bejleri, University of FloridaShow Abstract
XINGJING XU, University of Florida
Daniel Brown, University of Florida
Nithin Agarwal, University of Florida
Siva Srinivasan, University of Florida
GIS street centerlines are a staple of any geospatial transportation database. Many transportation studies, especially those looking at roadway safety, require access to locations and characteristics of horizontal GIS curves, given that curves represent potentially dangerous locations prone to traffic crashes. Despite these critical needs, curve inventories are surprisingly not typically available as part of standard GIS street databases. Manual delineation of curves is a time-consuming process that is not practical for large street networks with hundreds of thousands of roadways. Therefore, an automated procedure is necessary to identify the curves and calculate their attributes. This study, drawing upon the literature, develops and implements an improved method and a tool for automatic curve identification and calculation of curve attributes which among others include curve type, radius, length, and transition type. The study used the GIS streets basemap of the entire State of Florida as the data source. The research team validated the results by applying a combination of goodness of fit metrics, visual inspection and comparison with existing GIS curves. The results show that the improved method identifies the curves successfully, calculates curve attributes accurately, and executes rapidly. The improved method overcomes some of the shortcomings of previous studies and enables broader applicability to any GIS network. Unique improvements, not previously used in the literature, include detection of the spiral transition, the ability to detect curves on large complex dual-centerline GIS networks, and improved curve detection accuracy by utilizing new variables such as roadway speed and centerline vertex density.
Extracting Horizontal Curvature Data from GIS Maps: A Clustering Method
Bekir Bartin, Altinbas UniversityShow Abstract
Kaan Ozbay, New York University
CHUAN XU, New York University
This paper presents the use of a clustering method for automatically estimating horizontal curvature data and crash modification factors (CMFs) using GIS roadway shapefiles. The clustering method identifies distinct sections on a roadway, either curved or tangent, based on the proximity of the approximated curvature values of data points from GIS roadway centerline shapefiles, calculates horizontal curvature data and the corresponding CMFs. The results of the clustering method are compared with two other methods: (1) the mobile access vehicle method based on field GPS measurements and (2) the manual data extraction method based on satellite images. The comparison was conducted on a total of 24.7 miles of four NJ rural two-lane roads. The results showed that the CMFs estimated by the clustering method were within 12.2 and 15.5 percent of the ones produced by the mobile asset vehicle and the manual data extraction method, respectively. In addition, the sensitivity of the manually extracted horizontal curvature data was examined by conducting three additional independent trials. The average percent difference in the calculated CMFs between trials was 15.5 percent. This study therefore concludes that the clustering method can produce CMF estimates as accurate as the two other methods method much more efficiently in terms of time and money.
A Framework for Integrating and Conflating TIGER Data into an All Public Roads Network for the State of Florida
Brittany Wood, HDRShow Abstract
Paul O'Rourke, Florida Department of Transportation
Mark Welsh, Florida Department of Transportation
In 2012, the Federal Highway Administration expanded the Highway Performance Monitoring System’s (HPMS) reporting requirements for State Departments of Transportation to submit a Linear Reference System (LRS) that includes all public roads. This is otherwise known as the ARNOLD requirement or All Roads Network of Linear Referenced Data. A comprehensive and specific framework to integrate and maintain all local roads with the current LRS is needed in order to meet the federal requirements. However, similar to most states, the Florida Department of Transportation (FDOT) does not currently collect information on its local roads. Florida currently maintains an LRS of approximately 47,000 roadway miles comprising roads that are of interest to the FDOT. To include all the public roadways in Florida would increase the mileage by nearly 100,000 miles, which poses an immense challenge for the state of Florida and other states who do not already maintain a public roads network. In this paper, a geospatial methodology and framework is proposed to create and maintain an all roads network using the Census’ TIGER road network. This includes methods to implement and maintain topology, integrate local roads with existing state maintained roads (conflation), methods to assign required attributes to local roads, and a yearly maintenance plan. This geospatial framework can easily be applied to other states as the foundation for meeting the baseline ARNOLD requirements.
A Global Assessment of Street-Network Sprawl
Chris Barrington-Leigh, McGill UniversityShow Abstract
Adam Millard-Ball, University of California, Santa Cruz
Disconnected urban street networks, which we call “street-network sprawl,” are strongly associated with increased vehicle travel, energy use and CO2 emissions, as shown by previous research in Europe and North America. In this paper, we provide the first systematic and globally commensurable measures of street-network sprawl based on graph-theoretic and geographic concepts. We compute these measures for the entire Earth at the highest possible resolution. We generate a summary scalar measure for street-network sprawl, the Street-Network Disconnectedness index (SNDi), as well as a data-driven multidimensional classification that identifies eight empirical street-network types that span the spectrum of connectivity, from gridiron to dendritic (tree-like) networks. Our qualitative validation shows that both the scalar measure and the multidimensional one are meaningfully comparable within and across countries, and successfully capture varied dimensions of walkability and urban development. We further show that in select high-income countries, our measures explain cross-sectional variation in household transportation decisions. We aggregate our measures to the scale of countries, cities, and smaller geographies and describe patterns in street-network sprawl around the world. Latin America, Japan, southern Europe and North Africa stand out for their low levels of street-network sprawl, while the highest levels are found in the United States, northern Europe, and south-east Asia.
LiDAR-Only Vehicle Localization Based on Map Generation
Zhaozheng Hu, Wuhan University of TechnologyShow Abstract
Qianwen Tao, Intelligent Transportation Systems Research Center
Gang Huang, Wuhan University of Technology
Hao Cai, Wuhan University of Technology
Xianglong Wang, Wuhan University of Technology
Accurate self-localization is crucial for intelligent vehicles. Currently, intelligent vehicles mostly rely on Integrated Navigation Systems (INS) for accurate localization, which suffers from the GPS blind problem. In this paper, we propose a LiDAR-only localization method for intelligent vehicles. The method consists of two steps, one step for constructing LiDAR map and the other for localization by referring to the pre-built map. In this paper, we construct the LiDAR map by using a serial of nodes from 3D LiDAR point clouds. And each node is described by scene features, 3D structure, and sensor pose. Specifically, scene features are extracted by using a novel distance-weighted projection (DWP) method from 3D point clouds. Based on the pre-built map, we propose a multi-scale strategy for accurate vehicle localization, including topological coarse localization, frame level localization by matching scene features with those from the map, and finally metric localization from a fast LiDAR cloud registration method by referring to the 3D structure in the map. The proposed method has been tested by using the public KITTI database and the actual data collected in the field with the developed prototyped intelligent vehicle. The results demonstrate that the proposed method can achieve about 10 centimeter localization accuracy by referring to the pre-built LiDAR-only map. And the results demonstrate that the proposed method has strong robustness in different types of LiDAR sensors and different environment.
A Novel Map Matching Algorithm for Fixed Sensor Data Based on Probe Sensor Data
Qi Cao, Southeast UniversityShow Abstract
Ren Gang, Southeast University
Li Dawei, Southeast University
Zhang Jifei, Southeast University
Hao Mingyang, Nagoya University
Map-matching can provide much useful traffic information by aligning the observed trajecotories of vehicles with the road network on a digital map. And it always plays a prerequisite role for many intelligent transport systems (ITS) applications. Unfortunately, the current map-matching approaches are almost developed for GPS trajectories produced by probe sensors installed in the vehicles, and can’t deal with the trajectories recorded by the fixed sensor, such as camera, loops and mirowaves. Considering the huge application value provided by the fixed sensor data, in this paper, we propose a novel map-matching model called Fixed-MM, which is designed specislly for fixed sensor trajectories. Based on two key observstions from the real word data, the Fixed-MM considers (1) the utility of each route and (2) the travel time constraint to match the observed trip to the route with highest probabilities. At the same time, with the rules observed from the distribution of sample trajectories, the reasonable routes generation algorithom is also developed to generate the alternatives set for fixed-MM. Finally, the train datatset and the test dataset was extracted from both probe sensor dataset and fixed sensor data. And with the train dataset, the Fixed-MM was estimated and the estimated results of the parameters prove the rationality of the model. With the test dataset, the Fixed-MM was evaluated for the accuracy and efficiency, and the matching results show that the Fixed-MM can perfrom effectively and efficiently.
Computation of Horizontal Road Curvature from Sequential Video Log Images Using Adaptive Curve Registration
Zhaozheng Hu, Wuhan University of TechnologyShow Abstract
Mengchao Mu, Wuhan University of Technology
Yuting Li, Hebei University of Technology
Collecting road curvature data is crucial for both Road Asset Management Systems (RAMS) and Advanced Driver Assistance Systems (ADAS). This paper proposed a novel vision-based approach for horizontal road curvature computation. We first demonstrate that sufficient central angle is crucial for robust and accurate circle fitting from an arc segment. To achieve this goal, we adaptively use a number of sequential images for curvature computation to meet sufficient central angle requirement. And we proposed a modified Iterative Closest Point (ICP) method to adaptively register curves reconstructed from multiple video log images. Especially, the initial guess for ICP is derived from point correspondences by utilizing epipolar geometry constraint to ensure fast speed and avoid local minimal. Second, an error analysis model is applied to filter those reconstructed points with large reconstruction errors to further enhance curvature computation. Finally, road curvature is computed by fitting circle from the adaptively registered curves. The proposed method has been tested with the actual video log images collected on different curved road segments with the radii of 115m, 275m, and 430m, respectively. The results demonstrate that the proposed method is practical, reliable, and accurate to compute different degrees of road curvatures.
Online Map-Matching for Traffic Sensing on Highway Network with Call Detail Record Data
Yiming Wang, Beijing Jiaotong UniversityShow Abstract
Honghui Dong, Beijing Jiaotong University
Limin Jia, Beijing Jiaotong University
Hanzhong Pan, Traffic Management Research Institute of the Ministry of Public Security
Hongtong Qiu, Traffic Management Research Institute of the Ministry of Public Security
Yong Qin, Beijing Jiaotong University
Mobile phone detection technology is considered to be promising for traffic data acquisition with the advantages of Call Detail Record (CDR) data: all-weather collection, wide coverage and low cost, which is a good supplement to the existing traffic sensing methods on highway network. Online map-matching with CDR data is a crucial step and a big challenge for the application of mobile phone in traffic sensing. In this paper, we propose an online map-matching algorithm framework for traffic sensing on highway network with CDR data. This algorithm can solve two problem which cause traditional GPS online map-matching algorithm failure: (1) the oscillation and drifting of CDR data, and (2) CDR trajectories near highway will be mismatched onto highway. Based on Hidden Markov Model (HMM), this algorithm can output the maximum likelihood path over the Markov chain for highway trajectories while no output for non-highway trajectories. We utilize 100 CDR trajectories on highway and 100 trajectories near highway to evaluate this algorithm.The experiment result shows that the accuracy of our algorithm can reach 90.5% with 259.2s output delay, which is viable for traffic sensing.
Coupling the National Performance Management Research Dataset and the Highway Performance Monitoring System Datasets on a Geo-Spatial Level
Darshan Pandit, University of MarylandShow Abstract
Kartik Kaushik, University of Maryland
Cinzia Cirillo, University of Maryland
Integration of various datasets is crucial given the emphasis placed on holistic reporting of performance measures of various variables related to road transportation by the Moving Ahead for Progress in the 21st Century (MAP-21) Act. None is more confounding than the merger of geospatial datasets, which is necessary for example, to combine vehicle travel time and volume information for road segments. Such a merged dataset is released through the National Performance Management Research Dataset (NPMRDS). The NPMRDS is supposed to exclusively cover the National Highway System (NHS) and Strategic Highway Network (STRAHNET) sub-selected from the Highway Performance Monitoring System (HPMS). However, one finds that the coverage is not perfect. There are not only many extra road segments included in the NPMRDS, but also some NHS/STRAHNET roads segments are not fully covered by corresponding NPMRDS segments. Further, one finds very little literature about the method Texas Transportation Institute (TTI) uses to orchestrate the conflation. Therefore, it was endeavored to create a conflation algorithm which might perform better. The benchmark for the proposed algorithm is the identification of the segments wrongly conflated during the creation of the NPMRDS geospatial dataset. The proposed methodology uses a combination of five measures of similarity between the HPMS and NPMRDS segments. The proposed method successfully identifies significant numbers of mismatched segments: about 5% excess NPMRDS segments, and about 3% HPMS segments without NPMRDS counterpart.