Skip to contentUnited States Department of Transportation - Federal Highway Administration FHWA Home
Research Home
Report
This report is an archived publication and may contain dated technical, contact, and link information
Publication Number: FHWA-HRT-10-073
Date: November 2010

Roadway Geometry and Inventory Trade Study for IntelliDriveSM Applications Applications

CHAPTER 5. ANALYSIS AND COMPARISON OF ROADWAY DATA SOURCES

This chapter analyzes and compares roadway geometry and inventory data sources based on the technical characteristics of the databases and on the operational and business processes of the data providers. Data sources are stratified by public versus commercial data providers, and where applicable, public sources are further stratified by Federal, State, and local data providers. The final subsection summarizes gaps between data needs for IntelliDrive and availability of sources.

TECHNICAL ANALYSIS

Coverage

Needs for IntelliDrive

Successful deployment of IntelliDrive will require both a geospatial roadway network that provides the location and basic roadway geometry for all public roads and roadway attribute data that can be linked to the geospatial roadway network to support specific IntelliDrive applications. The applications will require complete national coverage.

Data Source Suitability

Commercial Databases: All five of the commercial roadway databases investigated in this study have a geospatial roadway network that includes all (or nearly all) public roads for the entire United States, as well as the southern (populated) Canadian Provinces. The commercial roadway networks include roadway attributes needed for vehicle navigation (one-way streets, turn restrictions, road classification, street name, route numbers, and address ranges).

National Public Databases:

  • TIGER/Line®: The Census Bureau TIGER/Line® database contains a geospatial roadway network that includes all (or nearly all) public roads for the entire United States, as well as some private roads that lead to occupied housing units. Roadway segments include attributes that identify address ranges and jurisdictional boundaries for defining and aggregating Census geography.
  • HPMS: HPMS is not a geospatial roadway network. It is a national inventory of roadway condition and performance data compiled and maintained by FHWA from data submitted by State DOTs in response to federally mandated reporting requirements. HPMS includes limited segment-specific attributes for all Federal-aid roads (e.g., number of lanes) and additional data on a sample of road segments selected based on highway functional classification.
  • NBI: NBI is not a geospatial roadway network. It is a national inventory of bridge locations and characteristics compiled and maintained by FHWA from data submitted by State DOTs in response to federally mandated reporting requirements. NBI includes all bridges with a span greater than 20 ft (6.1 m) that are open to the public.
  • HRCI: HRCI is not a geospatial roadway network. It is a national inventory of highway-railroad crossings compiled and maintained by the U.S. DOT from data submitted by State DOTs (and railroads) in response to federally mandated reporting requirements. HRCI includes all public and private railroad crossings, both at-grade as well as grade-separated (i.e., railroad bridges over roadways).

Statewide Public Databases: Most State DOTs have developed geospatial roadway networks, primarily to display State roadway inventory data that they collect. State roadway networks rarely extend beyond the State's borders, and there is generally little coordination between States to ensure that the roadway networks between neighboring States are consistent with respect to locational accuracy, feature resolution, and attributes. Additionally, a significant number of State roadway networks exclude roads for which the State DOT has no administrative, maintenance, or Federal reporting responsibilities such as local roads, toll roads administered by a separate authority, and roads on Federal lands.

With the exception of those data items specified by Federal reporting requirements, the roadway attribute data collected and maintained by State DOTs vary considerably from one State to another. In general, State DOTs maintain roadway inventory data only on those roads for which they have administrative or maintenance responsibilities. This represents about 10–15 percent of the total public roadway centerline mileage in most States.

Local Public Databases: Very few local road agencies develop and maintain their own geospatial roadway networks. Local agencies that have their own GIS capability typically use roadway networks developed by the State DOT or the Census TIGER/Line® roadway database, or the agency purchases a road network database from a commercial roadway database developer. Local agencies that require vehicle routing and navigation capabilities (e.g., E911 services) rely almost exclusively on commercial roadway data sources.

Data collection at the local level is extremely variable, with most local agencies collecting little, if any, roadway attribute data. Local data collection rarely extends beyond the jurisdictional or administrative service area boundaries of the local roadway agency.

Network Connectivity

Needs for IntelliDrive

Several of the key roadway attributes required for IntelliDrive applications (e.g., locations of intersections, entrance and exit ramps on freeway interchanges, and the identification of overpasses or underpasses where road segments cross but do not intersect at-grade) are most efficiently represented through the structure of the roadway database itself. Geospatial roadway databases that incorporate the basic network structure into the design of the database are defined as having network connectivity or network topology. At a minimum, this means that roadway segments always end wherever one road physically intersects another (e.g., at-grade intersections, freeway on- and off-ramps) and that road segments that cross but do not physically intersect (e.g., overpasses) do not end at the crossover point. Network connectivity is also a prerequisite for vehicle navigation and routing.

Data Source Suitability

Commercial Databases: All of the commercial roadway databases currently have the necessary network connectivity as well as additional roadway attribute data (e.g., identification of one-way roads and turn restrictions) needed to fully support vehicle routing and navigation.

National Public Databases:

  • TIGER/Line®: The TIGER/Line® database is structured using planar topology, which means that all line segments (roads, rivers, railroads, etc.) end wherever they cross any other line segment even if they do not physically intersect. The TIGER/Line® database could potentially be enhanced to support IntelliDrive applications but would require a significant amount of effort to identify overpasses and underpasses, one-way streets, and additional administratively defined prohibited and permitted vehicle movements.
  • HPMS, NBI, and HRCI: These databases are not geospatial roadway networks, and network connectivity is not applicable.

Statewide Public Databases: Most State geospatial roadway networks currently lack network connectivity. This is because State roadway networks have been developed primarily to display roadway inventory data and only need to support one or more LRS consisting of defined routes and linear measurements along those routes. Like the TIGER/Line® roadway database, State geospatial roadway networks could be enhanced to support IntelliDrive applications, but such an enhancement would require significant effort to identify and correct missing network connections. Most State DOTs currently do not have the available resources nor the perceived need to undertake such an activity.

Local Public Databases: Very few local road agencies develop and maintain their own geospatial roadway network.

Feature Resolution

Needs for IntelliDrive

IntelliDrive stakeholders and prior research have indicated that geospatial roadway networks should have sufficient feature resolution to distinguish between individual travel lanes on a roadway and to provide lane-level attribute information (e.g., lane width, pavement markings, cross slope).

Data Source Suitability

None of the current geospatial roadway networks, either public domain or commercial, provide lane-level feature resolution for all roads. Virtually all of the current roadway networks represent roadway segments as the approximate centerline of the travel way or paved surface. This means that two-way roads with no center median or physical barrier are depicted as a single line in the roadway database, representing the approximate centerline of the paved surface. Divided highways with an unpaved median or physical barrier separating travel directions are typically depicted by two lines, with each line representing the approximate midpoint of all through lanes in each travel direction. Individual lanes are typically represented only when they are separated from the general travel lanes by a physical barrier or median (e.g., HOV lanes with their own on- and off-ramps or channelized turn lanes separated by a curb). The actual number of through lanes associated with a roadway segment is an attribute of the roadway segment. The following provides some additional information for each data source.

Commercial Databases:

  • ALK®: Roadway segments depicted as centerline of the travel way.
  • DeLorme: Roadway segments depicted as centerline of the travel way.
  • Google™: Roadway segments depicted as centerline of the travel way.
  • NAVTEQand Tele Atlas®: NAVTEQ™ and Tele Atlas® also depict roadway segments as centerlines of the travel way. However, both NAVTEQ™ and Tele Atlas® offer enhanced lane information for certain roadway segments—specifically, freeway interchanges and complex intersections with dedicated turn lanes. The enhanced lane information does not explicitly provide lane-level feature resolution. Instead, information is provided to enable drivers to navigate to the correct lane on a roadway segment in sufficient time to enter a turning lane. Enhanced lane information is available for all major highway interchanges and for arterial intersections in approximately 36 cities. Additional cities are being added at a rate of about 10 per year.

National Public Databases:

  • TIGER/Line®: Roadway segments depicted as centerline of the travel way.
  • HPMS, NBI, and HRCI: These databases are not geospatial roadway networks, and feature resolution is not applicable.

Statewide Public Databases: Most State DOTs currently have geospatial roadway networks that depict roadway segments as the centerline of the travel way. However, a few State DOTs have not significantly enhanced their roadway networks in more than a decade. Their geospatial roadway networks depict roadway segments as the centerline of the entire roadway right-of-way. This means that divided highways are represented by a single roadway centerline located approximately in the median of the divided highway and freeway interchanges are represented as a simple intersection point.

Local Public Databases: Very few local road agencies develop and maintain their own geospatial roadway network.

Geospatial Accuracy

Needs for IntelliDrive

Potential IntelliDrive applications require a level of geospatial accuracy in the roadway network database so that a vehicle can unambiguously locate what roadway it is traveling on, where on the roadway it is relative to specific roadway features (e.g., a curve, steep grade, or intersection), and ideally, what lane of the roadway it is in. Currently, standard GPS locational accuracy is approximately 32.8 ft (10 m) absolute (circular error) and approximately 3.28 ft (1 m) with differential correction (as provided through the NDGPS program). This suggests that geospatial roadway networks necessary to support IntelliDrive should have a geospatial accuracy of no worse than 3.28 ft (1 m) absolute error.

Data Source Suitability

Current geospatial roadway networks have a geospatial accuracy of approximately 16.4–49.2 ft (5–15 m) absolute error, depending on how and when they were constructed. Most current geospatial roadway networks were initially digitized from digital orthoimagery. The corresponding accuracy of the roadway centerline depends on both the resolution of the source imagery and the accuracy of the digitizing technician to manually trace the centerline of the roadway from the imagery.

An alternative method is to drive each road in a GPS-equipped vehicle (with differential correction) and trace the vehicle's GPS position (as recorded every 1–5 s) to create or correct the position of the digitized roadway centerline vector. This method should be more accurate than manual digitizing from orthoimagery but is also subject to errors caused by forced lane changes along a roadway and by temporary loss of GPS signal due to urban canyons, tree cover, etc.

A third method, which has the potential to improve overall geospatial accuracy and produce centerline vectors for individual lanes, is to collect GPS tracks of a large number of vehicles traveling the same roadway over time. Using statistical averaging techniques, these GPS tracks can be aggregated to generate the centerlines of individual lanes along the roadway traversed by the vehicles.

Specific accuracies are presented for each geospatial roadway network data source.

Commercial Databases:

  • ALK®: 3.28–9.84 ft (1–3 m) (absolute error) on most major roads, no worse than 24.6 ft (7.5 m) on other roads.
  • DeLorme: 3.28–16.4 ft (1–5 m) in most urban areas, no worse than 24.6 ft (7.5 m) on other roads.
  • Google™: Unknown accuracy, appears to be no worse than 24.6 ft (7.5 m) on most roads.
  • NAVTEQ™: 3.28–16.4 ft (1–5 m) on all ADAS-compliant roads, 16.4–49.2 ft (5–15 m) on other roads.
  • Tele Atlas®: 3.28–16.4 ft (1–5 m) on all ADAS-compliant roads, 16.4–42.6 ft (5–13 m) on other roads.

National Public Databases:

  • TIGER/Line®: 24.6 ft (7.5 m) maximum absolute error, nationwide. The Census Bureau established this as its standard for acceptable positional error for all roadway geometry submitted by TIGER/Line® enhancement contractors and public agencies.
  • HPMS, NBI, and HRCI: The locational accuracies of roadway features in these national inventories are variable, depending on the data collection methods and accuracy criteria utilized by each State DOT that submitted the data.

Statewide Public Databases: The roadway networks developed by State DOTs vary significantly with respect to locational accuracy. Over the past decade, many State DOTs have initiated programs to improve the locational accuracy of their roadway network databases, utilizing high resolution (i.e., 1:10,000 scale or better) digital orthoimagery or GPS tracks collected by roadway inventory vehicles. The geospatial roadway networks produced from these efforts have locational accuracies of 16.4–32.8 ft (5–10 m) absolute error or better. About 60 percent of the State DOTs currently have or are developing geospatial roadway networks at this level of locational accuracy.

Local Public Databases: Very few local road agencies develop and maintain their own geospatial roadway network.

Attributes

Needs for IntelliDrive

The tables in chapters 3 and 4 provided detailed information on the specific roadway geometry and inventory attributes identified by IntelliDrive stakeholders with respect to availability by data source and to current and potential availability given emerging data collection technologies. This chapter provides more general observations regarding the consistency, measurement accuracies, and timeliness of roadway attribute data on a nationwide basis.

Data Source Suitability

Commercial Databases: All of the commercial roadway databases provide consistent feature and attribute definitions, attribute domains, measurement methods, and verification procedures across their entire roadway database.

Commercial roadway network developers also take a proactive role in verifying and updating the roadway inventory data included in their databases. Commercial database developers rely on multiple sources, including ongoing contact with State DOTs and local roadway agencies, feedback and error reporting from database customers, and in some cases, field teams that independently verify attribute changes reported from other sources.

National Public Databases:

  • TIGER/Line®: The TIGER/Line® database provides consistent and well-defined feature and attribute definitions, attribute domains, and measurement methods. However, other than the geospatial representation of the roadway network and locations of certain roadway features (intersections, freeway ramps, and railroad crossings), it does not include any of the roadway attributes identified by IntelliDrive stakeholders.
  • HPMS, NBI, and HRCI: These national inventory databases provide consistent feature and attribute definitions, attribute domains, and measurement methods. However, other than consistency checks on domain values, little if any of the data currently submitted by State DOTs to the national inventory databases is independently verified by the U.S. DOT for accuracy. Additionally, although each of the national inventory databases requires an annual submittal, attributes for specific features may be updated much less frequently. For example, bridge inspections are required every 2 years (4 years under certain conditions). Consequently, changes in some bridge attributes (e.g., clearance height) may not be reflected in NBI for up to 4 years after they occur.

Statewide Public Databases: Roadway attribute data maintained by State DOTs vary considerably from one agency to another with respect to attribute definitions, domain values, positional accuracy and resolution, data collection methods, and frequency of updates. Other than those data collection and reporting standards associated with the national inventory databases, there are no federally mandated standards for collecting and maintaining any of the roadway geometry or inventory data items identified by IntelliDrive stakeholders. Consequently, compilation of roadway attributes from individual State DOTs would require either the promulgation of additional national standards for data collection and reporting or substantial effort to document and reconcile differences across all contributing agencies.

Local Public Databases: Roadway attribute data collected and maintained by local roadway agencies vary even more than among State DOTs with respect to attribute definitions, domain values, positional accuracy and resolution, data collection methods, and frequency of updates. At the local level, there are no federally mandated standards for collecting and maintaining any of the roadway geometry or inventory data items identified by IntelliDrive stakeholders.

Data Format and Size

Needs for IntelliDrive

Roadway geometry and inventory data to support IntelliDrive applications must be stored in a format that can be read and utilized by a vehicle's on-board processors and can be efficiently transmitted and updated for specific (small) geographic areas and attributes.

Data Source Suitability

Nearly all of the geospatial roadway networks can be exchanged in one or more standard commercial GIS formats such as ESRI shapefiles, geodatabase files, or MapInfo TAB files. These files can be translated, with relatively little loss of information, into most commercial GIS software packages. However, transmitting and merging geographic subsets of the database requires additional processing by the receiving GIS software.

Commercial Databases: Standard distribution formats used by commercial database developers are as follows:

  • ALK®: Shapefile.
  • DeLorme: Shapefile, geodatabase.
  • Google™: Google™ does not distribute its Google Maps™ roadway network as a separate commercial product.
  • NAVTEQ™: Shapefile, TAB file.
  • Tele Atlas®: Shapefile.

NAVTEQ™ and Tele Atlas® also distribute their roadway geometry and attribute data using GDF, a European geographic data format that has become the international de facto exchange standard for navigation databases. While this ultimately may not be the most efficient format for transmitting roadway data needed for IntelliDrive applications, it has proven to be an operationally practical format for in-vehicle roadway navigation databases.

National Public Databases:

  • TIGER/Line®: Shapefile
  • HPMS: HPMS databases are collected and distributed as standard ASCII files, with each data record representing a roadway segment. Each record includes fields for a route identifier and beginning and ending mile point measurements needed to link the record to a State geospatial roadway network using a State DOT-specified LRS.
  • NBI and HRCI: Both of these national inventory databases are distributed as geospatial point features in shapefile format.

Statewide Public Databases: State geospatial roadway networks are typically stored and distributed as a geospatial database using the format of the predominant GIS software used by the agency (generally shapefile or geodatabase format). Roadway attribute data are usually linked to the geospatial roadway network using one or more LRS developed and maintained by the State DOT.

Transmittal of roadway attribute data stored using an LRS is significantly more complex than the transmittal of a geospatial feature database. It requires the receiving geospatial roadway network to have the route identifiers and linear measures for each LRS used to link a roadway attribute to the network. Additionally, because State DOTs continually update their LRS, these updates also need to be transmitted on a regular basis.

Local Public Databases: Very few local roadway agencies develop and maintain their own geospatial roadway network.

Methods and Frequency of Updates

Needs for IntelliDrive

IntelliDrive applications will require that roadway attribute updates that are critical to vehicle safety (e.g., bridge height and weight clearances, one-way street designations, turn prohibitions) be transmitted and incorporated into the vehicle roadway database soon after they are implemented. Failure to do so will, at best, undermine driver confidence in specific IntelliDrive applications and, in the worst case, contribute to a safety-related incident.

Data Source Suitability

None of the geospatial roadway networks or roadway inventory databases investigated in this study are updated and disseminated in real time.

Commercial Databases: Commercial roadway databases obtain updates on a continual basis from multiple sources, including regular contact with State DOTs and local road agencies, feedback from database users reporting corrections and changes in attribute information, data from field teams, and automated customer-generated content such a GPS tracks from vehicles and personal navigation devices (PNDs). Data from these multiple sources are verified, cross-checked, and reconciled with conflicting data and incorporated into a master internal roadway database.

Commercial database customers can obtain published updates at regular intervals ranging from as frequently as once a month to once a year, depending on the needs of the customer and volume of demand. Current publication frequencies of commercial roadway databases are as follows:

  • ALK®: Quarterly release of complete roadway network; monthly maintenance updates for some commercial customers.
  • DeLorme: Annual release of complete roadway database; semiannual release for some commercial customers.
  • Google™: Google™ does not distribute its Google Maps™ roadway network as a separate commercial product.
  • NAVTEQ™: Release of complete roadway network at least quarterly.
  • Tele Atlas®: Release of complete roadway network at least quarterly.

National Public Databases:

  • TIGER/Line®: A new version of the TIGER/Line® database was published annually during the TIGER/Line® accuracy enhancement program. The Census Bureau plans to continue to publish updates on an annual cycle, subject to available funding.”
  • HPMS, NBI and HRCI: The national inventory databases receive updates from State DOTs on an annual basis, but some of the data items reported by the States are based on periodic inspections or data collection activities that take place on 2-, 3-, or 4 year cycles. Additionally, there is often a significant lag between the time that data are submitted to the U.S. DOT by the States and the time the data are actually published. The lag time for HPMS data, for example, has historically been 6–9 months.

Statewide Public Databases: Other than those data items that are submitted annually to the U.S. DOT in compliance with Federal reporting requirements, most State DOTs have no regular schedule for collecting or disseminating roadway attribute data. Generally, roadway data collection efforts are determined by available funding and other programmatic priorities that divert staff and resources.

Local Public Databases: Roadway data collection among local roadway agencies is even more variable than among State DOTs. Resources are typically more scarce, and there are no specifically mandated roadway data collection requirements for local agencies. Many local data collection initiatives are one-time efforts with no specific plans for updates once the initial data are collected.

MAINTENANCE AND OPERATIONS ANALYSIS

Data Collection and Updating Technologies

Needs for IntelliDrive

Ideally, roadway geometry and inventory data used in support of IntelliDrive applications should be collected using methods that are consistent, meet the required levels of positional and measurement accuracy, and are amenable to efficient updating.

Alternative Data Collection Technologies

Most of the sources of roadway geometry and inventory data investigated in this study use similar technologies for collecting and updating their databases. The method currently used for most primary roadway data collection is to drive each road using a vehicle equipped with a differential GPS recorder and various data capture technologies such as videolog cameras, pavement condition measurement equipment, vehicle telemetry recorders, and increasingly, mobile LIDAR. By synchronizing all of the on-board data measurements to the vehicle location, it is possible to collect large volumes of roadway data simultaneously with one pass of the vehicle along the roadway.

In theory, nearly all of the roadway inventory data items identified by IntelliDrive stakeholders could be captured and measured using mobile data collection vehicles. In practice, however, the level of effort currently required to manually process videolog or mobile LIDAR imagery in order to identify, locate, and obtain key measurements of specific roadway features typically forces the sponsors of these data collection efforts to prioritize what features they extract and measure as well as what roads they collect data on.

A second data collection method that is useful in capturing basic roadway alignment and some roadway features (e.g., number of lanes, bridge and median locations, pavement markings) is airborne imagery. The primary advantage of airborne imagery is that it can capture data over large areas more quickly and at a lower overall cost per area than mobile roadway vehicles. Additionally, the collection of airborne imagery is often conducted as a cooperative project that is jointly funded by several State and Federal agencies, further lowering the cost to each agency.

Airborne imagery, specifically orthophotography, is the original source for most of the geospatial roadway networks currently in use, both public and commercial. It continues to be a major source for updating and correcting geospatial roadway network databases with respect to general alignment and incorporation of new roadways, particularly among several commercial roadway database developers. One reason for this is that a significant amount of orthoimagery is in the public domain, making it a free data source that commercial developers can use to improve their products.

An emerging data collection technology closely related to airborne orthoimagery is the use of IFSAR imagery to obtain reasonably high-accuracy (3.28–9.84 ft (1–3 m) absolute, 0.984–3.28 ft (0.3–1 m) relative error) 3 D measurements of roadway centerline alignment and elevation. While potentially less accurate than alignment measurements collected using mobile LIDAR, IFSAR has the benefit of producing a consistent measurement of curvature and grade for all roads over large geographic areas (i.e., entire States or even CONUS). This approach could be used to collect roadway geometry measures on low-volume local roads, where mobile roadway vehicles are not likely to be deployed.

A third data collection method is to collect and process GPS vehicle tracks from users of in vehicle navigation devices and PNDs. By collecting large numbers of GPS tracks over the same roadways, it is possible to calculate the centerlines of individual lanes, locations of new roads, locations of stop bars at intersections, paths of vehicles through intersections, and average speeds on roadways more accurately than could be obtained from a single pass of a roadway data collection vehicle. This technology is not able to collect all of the roadway attribute data identified by IntelliDrive stakeholders and will produce measurements with varying levels of error depending on the number of GPS tracks collected. In other words, measurements will be more accurate on higher volume roads with larger numbers of vehicles contributing GPS track information and less accurate on low-volume rural and local roads, where fewer GPS tracks are collected. However, GPS track data can also provide an historical record of changes in roadway traffic conditions over time (e.g., vehicle speeds, lane use) that could be used to develop behavioral profiles for specific roadway segments.

Data Source Suitability

Commercial Databases: The principal data collection and updating methods used by commercial roadway database developers are as follows:

  • ALK®: Airborne orthoimagery, user-generated GPS vehicle tracks.
  • DeLorme: Airborne orthoimagery.
  • Google™: Airborne orthoimagery, roadway videolog imagery.
  • NAVTEQ™: Airborne orthoimagery, roadway videolog imagery.
  • Tele Atlas®: Airborne orthoimagery, roadway videolog imagery, user-generated GPS vehicle tracks.

National Public Databases:

  • TIGER/Line®: Airborne orthoimagery.
  • HPMS, NBI and HRCI: National inventory databases are compiled from data submitted by State DOTs. No primary data collection is conducted by the U.S. DOT for these databases.

Statewide Public Databases: State DOTs collect roadway data using a variety of data collection technologies, including airborne orthoimagery and mobile roadway data collection vehicles equipped with differential GPS, videolog cameras, and pavement condition measurement equipment. Historically, much of the mobile roadway data collection has been in support of pavement condition monitoring. More recently, however, a number of State DOTs have initiated statewide data collection of roadway features (e.g., signs, guardrails, pavement markings) and are utilizing mobile roadway vehicles with video cameras and mobile LIDAR.

Local Public Databases: Local roadway agencies generally collect little or no roadway data on a regular basis. Any data that are collected at the local level are often done in cooperation with State DOT data collection activities.

Work Flow Processes

Needs for IntelliDrive

Roadway data to support IntelliDrive applications need to come from stable, reliable data sources with well-defined processes for data collection, storage, updating, quality control, and dissemination.

Data Source Suitability

In general, work flow processes are more clearly defined and better integrated among the commercial roadway database developers than among public agency sources. The reason for this is that, for commercial database developers, the roadway database is a marketable product that represents a primary source of income. The better the product, the higher the market share and greater the revenue. For public agencies, data collection is conducted either in response to a statutory reporting requirement or to support specific internal applications.

Commercial Databases: All of the commercial roadway database developers have established work flow procedures for collecting, verifying, updating, and disseminating each of the data items included in their roadway network databases. The specific processes differ slightly for each database developer.

Each of the commercial roadway database developers employs an in-house staff of GIS technicians to input, verify, and edit geospatial and attribute data received from various primary sources. These sources include publicly obtained roadway vector and orthoimagery, updates from State DOTs and local roadway agencies, error reports from database users, and other user-generated content such as GPS vehicle tracks. The new data are checked, verified, reconciled against other data sources, and incorporated into the master roadway network database on a continuing basis. On a periodic basis, a copy of the master roadway network is produced and published as a new version of the commercial database.

In addition to in-house database technicians, some commercial database developers deploy local field staff who work with State and local transportation agencies throughout the United States to obtain updated information on new road construction, changes in roadway alignments or traffic regulations, and temporary road closings due to construction. The local field staff also use mobile roadway data collection vehicles to verify changes reported from alternative sources and to collect new attribute data (e.g., street names, number of lanes, sign messages) that may not have been provided.

Methods used by commercial database developers to collect, verify, and update roadway databases are as follows:

  • ALK®: Ongoing contact with State and local roadway agencies, public domain data sources, feedback from database customers, user-generated content (GPS vehicle tracks).
  • DeLorme: Ongoing contact with State and local roadway agencies, public domain data sources, feedback from database customers.
  • Google™: Ongoing contact with State and local roadway agencies, public domain data sources, feedback from database customers, mobile field teams. (Note: this information is unverified.)
  • NAVTEQ™: Ongoing contact with State and local roadway agencies, public domain data sources, feedback from database customers, mobile field teams.
  • Tele Atlas®: Ongoing contact with State and local roadway agencies, public domain data sources, feedback from database customers, mobile field teams, user-generated content (GPS vehicle tracks).

National Public Databases:

  • TIGER/Line®: The Census Bureau maintains the TIGER/Line® database with an in-house staff of GIS technicians, who respond to feedback from users regarding errors in feature location, address ranges, and other attribute values (e.g., wrong street name). Additional feedback is provided by Census takers for both the decennial Census and the annual sample of respondents for the American Community Survey. Major updates and enhancements to the TIGER/Line® database are conducted through outside contracts.
  • HPMS, NBI and HRCI: National inventory databases are compiled from data submitted by individual State DOTs subject to a standard data reporting format. U.S. DOT staff conduct consistency checks to ensure that attributes values are within acceptable domain ranges and that summary statistics are reasonable (e.g., Interstates represent 0.5–1.5 percent of total road mileage in the State). However, the U.S. DOT does not conduct independent verification of attributes reported by State DOTs for specific road segments.

Statewide Public Databases: Among State DOTs, collection of specific roadway data items has historically been compartmentalized, with each operating division collecting the data it needs for its own internal applications (e.g., the Traffic Division collects traffic counts, the Pavement Division collects pavement data, and individual maintenance districts collect roadside features (guardrails, shoulders, signs) on roads within their geographic area of responsibility). The result has been an assortment of "legacy" databases scattered throughout the agency, using different data formats, storage media, data collection and updating procedures, quality control standards, and location referencing methods.

Over the past two decades, many State DOTs have tried to develop enterprise data repositories to facilitate data sharing across organizational units, using geographic location as the common identifier for roadway features and attributes. The success of these efforts has been mixed. In general, most State DOTs have been successful in developing a statewide geospatial roadway network and one or more LRS methods to enable them to display roadway inventory data stored in different legacy databases on the roadway network. However, relatively few State DOTs have successfully implemented agency-wide standards for data collection, including database documentation, quality control/quality assurance criteria, update or reverification frequency, or even storage media. As a result, roadway data collection at many State DOTs remains highly decentralized with minimal agency-wide coordination of data collection processes.

With respect to data collection for Federal reporting, many State DOTs have established separate groups within their organizational structure whose sole purpose is to compile and/or collect data for HPMS reporting. HPMS data items are compiled from databases maintained by individual organizational units, translated into the formats required for HPMS reporting, and submitted to FHWA on an annual basis. Some HPMS sample data items, such as roadway curvature, passing sight distance, or number of signalized intersections, are collected only for the designated HPMS sample sections and are not used by any other organizational unit within the agency.

Local Public Databases: Local roadway agencies collect roadway data on a sporadic basis, if at all. Work flow processes for data collection, verification, and updating are extremely limited or nonexistent.

Business Models

Needs for IntelliDrive

Ideally, roadway geometry and inventory data sources should recognize IntelliDrive stakeholders as potential customers and develop either new data products or enhancements to their current data products to address specific IntelliDrive application needs. At a minimum, the business models should have the flexibility to accommodate new data collection required for IntelliDrive.

Data Source Suitability

There is a clear dichotomy between public and commercial data sources with respect to who is viewed as the primary customer for the roadway data that they collect. This fundamental difference in business models between the public and private sector significantly influences how much each data source is likely to support IntelliDrive.

Commercial Databases: For commercial roadway database developers, the roadway data are either the primary product or are an integral part of the product or service that the commercial developer sells. The content and quality (i.e., geospatial accuracy, frequency of updates) of the data are tailored to meet the perceived needs of the target market. Details of each product are as follows:

  • ALK®: ALK®'s primary markets are the commercial transportation industry and personal navigation applications for mobile phones. Many enhancements to ALK®'s roadway network database are focused on data items of interest to commercial trucking, including height and weight limits on bridges, roadway restrictions, and truck-based services and POIs. ALK® also has implemented practices to ensure that key data items are updated and disseminated quickly to commercial customers. ALK® currently does not plan to develop roadway map databases specifically for the IntelliDrive application market. However, ALK Digital Maps™ could be extended to incorporate new roadway attributes should a supporting business case for this market materialize.
  • DeLorme: DeLorme's primary markets are outdoor recreational and natural resource exploration. To serve this market, DeLorme has created a basic roadway map database, primarily from publicly available sources, and has focused its supplementary data enhancement on identifying and locating off-road features such as hiking trails, campsites, back country topography, and natural POIs. Given its current market niche, DeLorme currently does not plan to develop a roadway database with additional attributes and enhanced geospatial accuracy necessary to support IntelliDrive applications.
  • Google™: Google™'s target market for Google Maps™ appears to be location-based services marketing to Internet and mobile phone users. Google Maps™ and the associated roadway map database that Google™ recently developed are provided as free services to users of Google™'s Web site and search engine. Google™ has supplemented the Google Maps™ roadway database with visual enhancements like "Street View," additional public domain data like transit stops, and expanding its POI database. All of these enhancements seem to be directed at making Google Maps™ the most popular Internet map search site, which in turn, increases the volume of searches for location-based businesses through Google™. At the time this report was written, Google™ did not appear to have any interest in marketing Google Maps™ as a separate product or in enhancing Google Maps™ to meet the requirements for IntelliDrive applications.
  • NAVTEQ™ and Tele Atlas®: The principal markets for NAVTEQ™ and Tele Atlas® are vehicle and equipment manufacturers for in-vehicle navigation systems, manufacturers of PNDs like Garmin or TomTom, mobile phone manufacturers like Nokia, and public agencies and commercial clients who want a geospatially accurate and routable roadway network for various applications. Both developers place a high priority on updating navigation-related data items such as new streets, changes in one-way streets, etc. and employ large, geographically dispersed field staffs to proactively identify and verify roadway changes from multiple sources. Both NAVTEQ™ and Tele Atlas® approach IntelliDrive as a potential new market area and already have begun making enhancements to their current roadway data products in anticipation of supporting future IntelliDrive requirements.

National Public Databases:

  • TIGER/Line®: The Census Bureau developed the TIGER/Line® database to enable it to define and update nationwide Census geography (e.g., Census Blocks, Block Groups, and Tracts) consistently and efficiently based on physical linear features such as roads, railroads, and rivers. Additionally, the TIGER/Line® road network was developed to help Census-takers locate housing units based on address ranges. The widespread popularity and use of the TIGER/Line® database for other GIS applications is recognized and encouraged by the Census Bureau. However, the collection and incorporation of new attributes into the TIGER/Line® database goes beyond the Census Bureau's mission and would have to be supported entirely by outside resources.
  • HPMS, NBI and HRCI: National data inventories exist to respond to Federal statutory requirements for periodic reports to Congress on critical elements of the Nation's transportation infrastructure. The national inventory databases provide a consistent, well-defined set of selected roadway attributes, some of which may satisfy IntelliDrive application needs. However, Federal reporting requirements rely almost entirely on State DOTs for data accuracy and quality control. Federal agencies conduct little or no independent verification of the accuracy of individual data items reported by the States. Additionally, the frequency with which Federal inventory databases are updated is not compatible with IntelliDrive requirements for most data items.

Statewide Public Databases: State DOTs collect roadway data for one of two reasons: (1) to meet federally mandated reporting requirements or (2) to support internal business needs for roadway construction, maintenance, or operations. Data collected by State DOTs in response to Federal reporting requirements are typically not utilized directly by the State DOT for its own business needs. In contrast, data that are used to support specific agency operational or maintenance needs are often collected, stored, updated, and administered by individual organizational units, with little or no coordination among organizational units.

Local Public Databases: Local roadway agencies generally collect little or no roadway data on a regular basis. Any data that are collected at the local level vary considerably from one agency to another, even within the same State.

CHALLENGES AND OBSTACLES TO PROVIDING ROADWAY GEOMETRY AND INVENTORY DATA FOR INTELLIDRIVE APPLICATIONS

The trade study identified several current data sources and promising new data collection technologies that can produce most of the roadway geometry and inventory data items identified by stakeholders as necessary or highly desirable for potential IntelliDrive applications. However, the study also identified a number of current data gaps and challenges that must be addressed in order to provide the necessary data coverage, level of detail, and timeliness of updates to support IntelliDrive applications on a nationwide, operational basis. These data gaps and challenges are summarized in this section.

Variability in Coverage for Roadway Data

The investigation of current sources of roadway geometry and inventory data revealed a clear and consistent hierarchy for collecting roadway data based on highway functional classification and geography. In general, the highest quality and most extensive data are available for the highest functional class roads—interstates, freeways and other limited access highways, and other principal arterials. These roads account for approximately 5 percent of the total roadway centerline miles in the United States but carry more than 50 percent of total annual vehicle miles of travel (VMT).(3)

In contrast, data on local roads (including minor rural collectors) are currently limited to basic roadway geometry and navigation information (e.g., turn restrictions, intersection locations) provided in commercial roadway network databases. Local roads account for over 75 percent of total roadway centerline miles but carry just over 15 percent of total annual VMT.

There are currently few, if any, incentives for either commercial roadway database developers or public transportation agencies to significantly expand their data collection on local roads. For commercial database developers, it is a matter of economics. The overwhelming majority of customers for roadway navigation data are traveling on nonlocal roads and want current and accurate data for those roads. Consequently, roadway database developers have traditionally focused their data collection activities on updating nonlocal roads. In order for them to provide equivalent levels of data collection on all local roads, they would have to expand their field staffs significantly. Similarly, because local roads are not eligible for most Federal-aid road funding, there are few incentives for State DOTs to collect and maintain roadway inventory data on these roads. Even HPMS requires only limited summary information on non-Federal-aid roads.

The administrative and maintenance responsibilities for local roads typically fall on local governments (counties, cities, or towns). The amount of data collected by these local agencies varies significantly, with no standards and little guidance provided about what data should be collected, data definitions, data formats, or even storage media. Some agencies collect extensive, high-quality data that could be used directly to populate a roadway database; other agencies collect no roadway data at all or maintain their data as paper records. Given this wide variation in local road data, it may actually be more efficient to implement a nationwide data collection effort for key local road data than to try to coordinate and standardize roadway data collection across more than 3,000 counties and 20,000 cities and towns.

Given the realities of current roadway inventory data collection in the United States, IntelliDrive stakeholders need to address the following two questions:

  • What is the absolute minimum roadway geometry and inventory data that is required on all U.S. roads in order to implement early IntelliDrive applications?
  • Can early IntelliDrive deployments function successfully if some roadway inventory data are only provided for higher functional class roads?

If the answers to these questions indicate that additional data need to be collected for local roads before early IntelliDrive applications can be deployed, then a major, nationwide data collection effort may need to be undertaken to acquire the minimum necessary roadway data items.

Resolution of Roadway Features

IntelliDrive stakeholders identified several roadway attributes that they would like to have for individual lanes, including lane width, cross slope, posted speeds, and longitudinal pavement markings. However, none of the current geospatial roadway networks, public domain or commercial, provide lane-level feature resolution, except where lanes are separated by a physical barrier or unpaved median.

Two of the commercial roadway database developers, NAVTEQ™ and Tele Atlas®, offer "enhanced lane information" for certain types of roadways—specifically, freeway interchanges and complex intersections with dedicated turn lanes in selected urban areas. This enhanced lane information does not produce separate geospatial features for each lane but does include additional data items associated with the roadway segment to identify specific lanes on the roadway segment and relevant attributes associated with each identified lane. Current lane-level attributes include turn restrictions and pavement markings but not the specific location of the lane centerline itself.

Both NAVTEQ™ and Tele Atlas® are collecting and incorporating enhanced lane information into their roadway geometry databases in preparation to support future ADAS applications. IntelliDrive stakeholders and application developers need to evaluate whether the architecture for lane-level data offered in these commercial databases can meet the requirements for proposed IntelliDrive applications or whether new geospatial roadway networks will have to be built to provide lane-level feature resolution.

Current Availability of Specific Roadway Attributes

Although nearly all of the roadway data items identified by IntelliDrive stakeholders could be collected using current or emerging data collection methods, many of these items are not currently being collected by commercial roadway database developers or public agencies as part of routine data collection activities. Specific data items currently not being collected include the following:

Roadway Geometry:

  • Vertical curvature (points of curvature and tangency, curve length).
  • Available sight distance (passing and stopping).
  • Cross slope/superelevation (roadway, lane, and shoulder).

Roadway Inventory:

  • Clear zone width.
  • Barrier and guardrail locations and characteristics.
  • Sidewalk locations.
  • Lane-level characteristics (width, function type, posted speeds).

Intersection Characteristics:

  • Unique intersection ID.
  • Detailed intersection geometry (location of corners, width of turn lanes).
  • Pavement markings (crosswalks, bicycle lanes, stop bars).
  • Vehicle paths through intersection.
  • Traffic signal characteristics (priority and preemption, precise location of signal head).

Other Geospatial Features:

  • Commercial vehicle facility characteristics (parking area capacity).

In the absence of new mandated reporting requirements or incentives, it is unlikely that these roadway data items will be collected by public agencies or commercial database developers. IntelliDrive stakeholders need to determine which of these data items are essential for specific applications and then identify strategies for collecting and maintaining each essential data item.

Data Standards

Many of the roadway geometry and inventory data items that are currently being collected do not have well-defined or widely applied standards for feature and attribute definitions, attribute domains and values, data formats, or positional accuracy. The only standards that currently exist on a nationwide basis are those promulgated by the U.S. DOT for the national roadway inventory databases (HPMS, NBI, and HRCI) and the exchange standard for roadway network databases as embodied in GDF, utilized by NAVTEQ™ and Tele Atlas®.

Roadway inventory data that are collected by State DOTs or local road agencies for their own internal applications but are not required by any of the national inventory databases can vary significantly with respect to definitions, attribute values, etc. from one agency to another. The level of effort required to assemble and reconcile data from these multiple sources is likely to be substantial.

For those data items where nationwide standards do exist, the current standards may not be sufficient to satisfy the requirements for IntelliDrive applications. For example, HPMS requires data on median, shoulder, and intersection characteristics to be reported only for selected sample sections and data on roadway curvature and grade to be collected only for broad ranges of values over the entire roadway segment rather than as specific measurements.

Data standards need to be developed to ensure that current and new roadway geometry and inventory data sources meet the needs of proposed IntelliDrive applications. The development of IntelliDrive data standards needs to be a cooperative effort involving IntelliDrive application developers and roadway database developers in both the public and private sectors.

Database Maintenance and Updating

IntelliDrive applications will require that updates to roadway data items be incorporated into the master roadway database quickly and accurately and be transmitted efficiently to in-vehicle databases.

Current reporting requirements and procedures for national inventory databases do not meet basic IntelliDrive requirements for accurate and timely updates. State DOTs are required to submit updates for national inventory data items once a year but may actually update attribute values associated with a specific feature according to an inspection schedule that is less frequent (e.g., 2–4 years for bridges). Furthermore, since the U.S. DOT does not conduct independent verification of data submitted by States DOTs, data errors are incorporated into the inventory database.

Commercial roadway database developers take a much more proactive approach to data updates, utilizing a variety of sources in addition to State DOT reports, including feedback from database customers and reports from field teams. Commercial master roadway databases are updated on a continuing basis, and changes in key roadway attributes are incorporated as soon as they are received and verified.

Regardless of how a roadway geometry and inventory database is initially developed, a long-term home must be found for it that will provide the database maintenance, updating, and verification functions necessary to meet IntelliDrive application requirements. IntelliDrive stakeholders need to determine whether these database maintenance functions can be achieved within a public agency structure such as the U.S. DOT or through a long-term maintenance contract with a commercial database firm or if overall responsibility for developing and maintaining IntelliDrive roadway geometry and inventory databases should rest entirely with the private sector.

ResearchFHWA
FHWA
United States Department of Transportation - Federal Highway Administration