APPENDIX B: WHITE PAPER ON NEW DATA SOURCES AND THEIR IMPACT ON ATM

INTRODUCTION

ATM relies upon data sources to provide a feedback loop into planning and operations for traffic management. This paper discusses new data sources for ATM and their impact on ATM planning, implementation, and operations. Traditional data sources used by public and private agencies alike include data collected through general hardware devices, which may include vehicle detectors, closed-circuit television (CCTV) cameras, or similar devices. The industry is leaning more on data sources that are nonintrusive, require limited operations and maintenance resources, and allow for greater geographic coverage than traditional agency sensor networks. An opportunity exists to improve both the quality and quantity of data collected and utilized by transportation agencies for advanced operating strategies such as ATM. These alternative data sources could improve not only the volume of data collected but also the speed at which it can be collected and analyzed. With the introduction of new data sources, an opportunity exists for improved implementation of ATM strategies such as adaptive traffic signal control, adaptive ramp metering, dynamic lane reversal, dynamic lane use control, and dynamic speed limits, among others.

STATE-OF-THE-PRACTICE

ATM has traditionally relied upon ITS field devices to provide data for planning and real-time operations, including dynamic and potentially automated strategies in some instances. These data sources are used by agencies to improve transportation systems; however, they require maintenance to remain accurate and are often limited in their ability to provide real-time information or provide performance metrics without processing. Data collected from current sources such as traffic signals, radar detectors, vehicle detectors, and video cameras may not be able to be stored for long periods of time due to a lack of storage space or agency policies. Older data can also be erased from the system to make space for new data. This, in turn, makes accessing previous data very difficult. Archived data are essential for agencies to make informed decisions and design choices when planning which ATM strategies need to be implemented. The extent or operating area for ATM strategies are often limited by the geographic coverage of agency-owned data collection infrastructure.

CONSIDERATIONS AND IMPACTS ON ATM

Data sources such as private sector data, connected vehicles/infrastructure, and other emerging data sources have a positive impact on ATM strategies. Traditional data sources such as loop detectors or CCTV surveillance images do not provide detailed insight or require operators to observe the data and perform an action. Data collected from emerging data sources, including private sector data, provide another alternative for transportation agencies when planning transportation systems at the local, state, or federal levels.

Types of new data sources include but are not limited to the following:

• Connected Travelers: User-experienced and user-inputted data include road conditions, traffic conditions, travel speeds, and road hazards. This data type also includes private big data sources such as Waze^© and INRIX^TM.

• CVs: Vehicles with the capability to communicate with other vehicles, vulnerable users such as bicyclists or pedestrians, or connected infrastructure provide additional data.
• Other Emerging Data Sources: Other data sources include infrastructure with the ability communicate and receive signals and notifications, high-definition maps, etc.

CONNECTED TRAVELERS INCLUDING PRIVATE SECTOR DATA

Data from connected travelers is typically collected directly from the traveler using dedicated short-range communications, Bluetooth connections, cellular networks, or Wi-Fi networks. Data are collected through the traveler’s mobile devices to provide information on travel speeds, travel times, routing, queues, etc. These data can be randomly collected from users using anonymous data collection measures or could include user-inputted data from mobile applications. Data are aggregated to provide segment or corridor conditions such as speeds, bottlenecks, and queue locations. Unlike agency sensors, these data are not able to provide lane-by-lane speeds and travel conditions. Although these data are subject to potential errors because it can be based on survey-like input from the public, this approach is one of the most cost-effective ways get near-real-time traffic data. The connected traveler data collected by agencies allows them to make more informed decisions regarding how to direct traffic, especially in emergency situations, without investing in major infrastructure. The data could also improve how agencies communicate with those traveling on affected roads, including transit vehicles.

CONNECTED VEHICLES

CVs are another emerging data source that could greatly benefit transportation agencies implementing ATM strategies because they provide more accurate information relating to driver and vehicle behaviors on the road, as well as location-specific information. CVs use wireless communications and the vehicle’s onboard computer system to communicate directly with infrastructure or other vehicles. CVs send out alerts to warn of potential hazards on the road or delays due to collisions or congestion. These alerts help drivers avoid certain roads and areas. They can also provide signal phase and timing information directly to the vehicle to optimize signal operations.

Three types of connected vehicle applications exist, including vehicle-to-vehicle (V2V), vehicle-to-other objects (V2X), and vehicle-to-infrastructure (V2I). V2V applications are mainly used for safety, V2X applications are used by vulnerable road users such as bicyclists and pedestrians, and V2I applications are used for traffic management. Commercial CVs are already utilized by private companies to collect data on real-time vehicle speeds on roadway links. Some private companies currently share these data with DOTs; however, it is possible for DOTs to collect these data themselves. If DOTs upgrade their technology and meet the communication requirements to use connected vehicle data sources, they will be able to collect these data themselves without needing to go through and pay private companies to collect these data for them (Gettman et al. 2016, 2018) A key challenge for agencies when considering how to collect, ingest, and use connected vehicle data is data overload, particularly for real-time operations. ATM systems and traffic operations staff could easily get overwhelmed with multiple data points coming in from multiple vehicles; these wide-ranging data points could include information on pavement/friction, vehicle braking, speed, queue, distance, etc.

OTHER EMERGING DATA SOURCES

Other examples of emerging data sources to support ATM include infrastructure-based equipment such as mobile sensors and high-definition maps. Data collected from mobile sensors such as light detection and ranging systems and 3D camera analytics provide additional data sources for ATM. These data could be very useful for agencies when implementing ATM strategies intended to safely restore traffic flow after traffic incidents. Data collected from high-definition maps, which is typically shared with public agencies by private companies, could potentially be useful for implementing V2I connected vehicle applications. Given the large quantity of intersections within one map alone, big data tools and technologies are necessary for traffic agencies to process and be able to use the data.

BENEFITS OF EMERGING DATA SOURCES

Emerging data sources can help improve a variety of ATM strategies including adaptive traffic signal control, adaptive ramp metering, dynamic lane reversal, dynamic lane use control, and dynamic speed limits. For example, they can help improve incident responses, identification of hazards and warnings on roads, speed warnings to drivers due to heavy traffic or dangerous weather conditions, accuracy of travel time messages, and hazard detection when traffic infrastructure equipment fails or is malfunctioning. Data can be available for a larger geographic area than the area supported by agency infrastructure, sensors, and communications networks. This expanded coverage is particularly relevant for rural corridors that often have little to no agency monitoring infrastructure in place.

Emerging data sources and tools provide many benefits to both public and private agencies. The use of emerging data sources instead of more traditional data sources (i.e., radar or video detection) provides agencies with higher volumes of data at a faster rate and allows agencies to keep a larger amount of data stored due to increased storage capacities. Although data such as connected vehicle data are not able to be stored in traditional storage locations for existing TMS due to its large size, the data can be stored in online cloud repositories.

Incorporating emerging data sources can also be beneficial because it allows archived data to be accessible for on-demand analytics. Being able to access archived data could be beneficial to agencies trying to predict travel behavior and would aid in actively managing traffic operations according to real-time traffic conditions. It is crucial for agencies to access real-time data to improve traffic operations and plan for any potential future infrastructure that addresses gaps in existing infrastructure or modes. Having access to real-time data as it occurs allows agencies to forego the waiting period usually needed to process traffic data.

Emerging data tools and technologies are necessary when handling data sources because existing TMS are not equipped to handle the large quantities of data. For example, given that the existing systems are not able to process the incoming large volume of data, a big data solution is to use a streaming data software system that can store data in the short term, process and analyze it, and then erase it from its short-term memory and move it into its long-term memory storage. Another improvement to existing systems that these tools and technologies can improve is the rate at which incoming data gets processed and analyzed. Big data tools and technologies are capable of parallel processing, which allows large volumes of data to be split into smaller portions and the information from the smaller portions to be processed at the same time. This processing can be performed either using a single machine or several machines linked together and working simultaneously. These and other improvements would be very useful when

processing big data from emerging data sources because data sometimes needs to be processed in real time. These processes allow data to be output directly to transportation agencies in the form of usable performance metrics and performance dashboards.

CASE STUDIES

ATM is reliant on data to provide the active component of ATM. As data sources have evolved from the typical brick and mortar data sources, agencies have used other types of data to inform traffic operations. Those data sources include connected vehicle data and data from connected travelers. The following case studies demonstrate the shift to using emerging data sources for ATM.

Statewide Traffic Incident Management in Pennsylvania

The Pennsylvania DOT (PennDOT) manages nearly 40,000 miles of roadway and operates four regional TMCs. They have used ITS infrastructure to monitor roadways for general operations and to identify and respond to incidents. Because the entire state is not outfitted with ITS infrastructure to the same degree, the gaps in infrastructure directly contributed to a lower level of active operations in specific geographic areas.

To provide consistent service across the state, PennDOT determined the most cost-effective solution would be to leverage existing contracts with private sector big data sources to fill gaps in ITS infrastructure coverage. PennDOT utilized INRIX^TM real-time traffic data as part of a license agreement with the I-95 Corridor Coalition’s Vehicle Probe Project and the Waze^© Connected Citizens Program. The PennDOT Road Condition Reporting System and Crash Reporting System were also used as data sources.

After ground-truthing the data to determine accuracy as compared to traditional detection methods, PennDOT integrated the private sector big data sources into the TMC operators’ workstation interface so they can receive alerts about congestion events or incidents directly from the big data sources. These alerts can help prevent secondary crashes; initial reporting showed that 65 percent of secondary crashes occurred without TMC operators being aware of the initial crash (INRIX^TM n.d.). The same study found that INRIX^TM detected over 81 percent of the initial crashes (INRIX^TM n.d.). These changes also resulted in more accurate DMS messaging for queue protection and faster responses by safety service patrols. PennDOT also implemented automated DMS messaging changes for queue clearance based on big data congestion information and virtual queue detection and smart work zone capabilities. Another change that occurred to operations as a direct result of using the private sector data sources related to TMC hours of operations; one TMC’s hours of operation were changed to reflect congestion patterns identified with private sector data.

Autonomous Shuttle Pilot in Utah

An autonomous vehicle is a vehicle capable of operating without a driver. An autonomous shuttle pilot was launched in the state of Utah in April 2019 and continued until September 2020. The shuttle utilized for this pilot program was leased from EasyMile^©—one of the few companies providing autonomous shuttles in the country.

The autonomous vehicle used in this project was a shuttle with the capability to transport up to 12 passengers at speeds up to 15 mph. Because the shuttle operates without a human driver,

it needs a preprogrammed route to follow, which requires planning and design by agencies. Because autonomous vehicles follow a predetermined route designed by traffic agencies, they can better predict and avoid potential areas of high traffic congestion.

This shuttle, as well as other autonomous vehicle types, are able to detect any objects nearby that can pose a threat to the safety of the passengers onboard. Some examples of this include the shuttle’s ability to detect other vehicles (in close proximity), bicyclists, pedestrians, and any potential hazards on the road. Due to their detection capabilities, automated vehicles can avoid potentially dangerous situations. In fact, this capability makes them safer and more likely to avoid collisions than traditional human-driven vehicles.

Throughout the duration of the project, a total of 6,878 riders were given the opportunity to try the autonomous shuttle. In a survey of 822 riders conducted after their rides, 98 percent of riders reported feeling safe during shuttle ride and 95 percent favored the use of autonomous vehicles more than they had before the ride (Utah DOT 2021a, 2021b).

This autonomous shuttle only operated at low speeds and in certain traffic conditions (e.g., daytime), suggesting that various improvements can be made. However, the Utah DOT learned valuable lessons about the types of data might be generated by an autonomous vehicle to support ATM strategies. In addition, some challenges identified during this project included a passenger being injured after the shuttle stopped very suddenly and the shuttle having to stop operations when maintenance was required. As more agencies push toward smart and connected cities, a greater possibility exists for improvements to CVs and infrastructure.

RELEVANCE TO KEY CONCEPTS

Table B-1 summarizes the relevance of new data sources to the major topic categories of the ATM guide. This matrix helped guide the inclusion of new and emerging data sources content in the final guide.

Table B-1. Relevance of New Data Sources to the ATM Guide Topic Categories.

MAJOR GAPS AND CHALLENGES

Emerging data sources are directly impacting ATM by providing a more data-rich environment and more options for agencies with limited funding. However, emerging data sources do have some challenges in their implementation. This section describes some of these challenges.

Staffing Needs and Capabilities

Staffing needs and capabilities are a challenge when implementing emerging data sources. Private companies have become acquainted with big data but, in most cases, the public sector has not. Staff time and funding are required to learn about big data. In addition, staff will need to be trained in how to use new types of data and how to integrate these data into ATM operations and decision-making processes. When resources are limited, it may be appropriate for agencies to outsource their work to independent contractors. Emerging and alternative data sources produce greater quantities of data that must be processed and analyzed to be used effectively. Agencies already have limited staff resources for data processing; processing greater quantities of data might be staff prohibitive.

Software/Technical Compatibility

When implementing new data sources for ATM strategies, agencies must account for their existing software and hardware capabilities. Existing equipment and software may not have the bandwidth to handle connected vehicle data sources and may require upgrades. Similarly, ATM operating processes need to be evaluated to see if it is feasible to rely on private sector data versus more traditional agency data sources. Transportation agencies need to have the proper technology to be able to collect, process, and store this information in addition to integrating these data into ATM operations. Although agencies may be currently using connected infrastructure to gather data, they may not have the capabilities to properly process these data. For example, existing advanced traffic management system client software may not be compatible with big data tools and technologies.

Security and Privacy

Security and privacy are at the forefront of projects involving emerging data sources. Data from emerging sources (e.g., data from CVs) is stored in an online cloud system, which poses cybersecurity threats from any potential hackers and is sometimes not allowed by IT administrators at transportation agencies (Khattak, et al. 2018). Potential cybersecurity risks involve people’s locations being tracked and their personal information being available to hackers (Khattak, et al. 2018). IT administrators must be a key stakeholder in the procurement and integration of emerging data sources to protect the security of the agency (Khattak, et al. 2018).

Coordination between Local Agencies and State DOTs

Coordination between local agencies and state DOTs or Federal agencies is necessary when ATM strategies cross agency borders. This coordination might be difficult when one agency is equipped with the tools and technology necessary to handle big and emerging data sources, but the other is not. This challenge will become less significant once emerging data

sources such as CVs and connected infrastructure are more widely used throughout the country. In some cases, one agency might have a license agreement with one private sector data provider while another agency might have a different agreement with another agency. These overlapping agreements may waste limited agency funds by duplicating data sources or may restrict agency partners from receiving the full benefits of the license agreement.

Cost

To implement the use of emerging data sources in transportation agency ATM strategies, existing software and systems must be upgraded to meet emerging data software requirements. Emerging data tools and technologies will also need to function within existing systems to be able to process, store, and analyze alternative data sources. In addition, IT staff for these agencies must know how to work with the new systems and be able to easily fix any issues that arise during operations. Meeting all of these needs may result in significant costs borne by the transportation agencies. Securing this funding is likely easier for larger agencies such as state DOTs but significantly more difficult for smaller, local public agencies.

KEY REFERENCES