PROJECT 1
SMART AI-TECHNOLOGY EMPLOYMENT FOR CRASH DATA ANALYSIS
PROJECT DESCRIPTION
Statistics from the National Highway Traffic Safety Administration [1], [2] show that the United States in 2022 there were 42,795 fatalities and about 2.5 million injuries resulting from motor vehicle traffic crashes. Among many crashes, pedestrian-related car crashes hold significant importance [3]–[5] due to their potential to cause severe injuries and loss of life, as well as their broader societal impact. These crashes underscore the vulnerability of pedestrians in collisions with vehicles. The consequences extend beyond individuals involved; the crash outcomes affect families and communities. Addressing pedestrian crashes requires a holistic approach that combines improved infrastructure, traffic regulations and enforcement, education efforts and public awareness campaigns, emergency / trauma medical care, and innovative vehicle safety technologies. The US DOT’s National Road Safety Strategy [6] and the Safe Systems Approach [7] reinforce the need to create more pedestrian-friendly environments and reduce the human and economic toll of these crashes, while fostering safer and more inclusive communities. In this regard, this research will take an initiative effort with crash narrative data – type of data that have not been exploited well historically [8]– to extract new insights about pedestrian-related vehicle crashes. Crash narratives include crash-related details, facilitating a deeper comprehension of each incident. By examining a collection of crash reports, one can discern recurring patterns and trends associated with specific attributes [9], such as particular human, roadway, vehicular, traffic control, or geographical factors. The primary objective of this research is to uncover new insights that could serve as fundamental stepping stone to foster advancements in traffic safety management. Moreover, this study aims to augment the existing knowledge base by creating an innovative methodology that harnesses Artificial Intelligence (AI) and Natural Language Processing (NLP) to efficiently delve into crash narratives, thus enhancing our level of understanding of such crashes. Methodological advancement and findings key to transportation safety will be incorporated into various educational and outreach programs at UNLV.
PRINCIPAL INVESTIGATOR(s)
Jee Park
Shashi Nambisan
Jeehee Lee
FUNDING SOURCE & AMOUNT
USDOT, 60,767 Federal, $47,548 Non-Federal
PROJECT START DATE
June 1, 2023
RESEARCH OUTPUT & IMPACTS
The successful completion of this project will offer several key values in transportation safety research and education. First, by replacing the time-consuming manual reading process with an automated approach based on NLP and AI, we will streamline operations and enhance efficiency in transportation safety analysis. Second, this initiative will significantly reduce (if not eliminate) the potential for errors and unreliable outcomes, ensuring the delivery of consistent and dependable results to analysts. Third, our improved understanding on crash factors can enhance decision-making at policy levels and aid in formulating effective strategies to enhance road safety, including targeted countermeasures. Last, this endeavor aligns with recent technological advancements and their practical implementation in advancing transportation safety, which can have a broader impact if applied to other key parameters of interest by safety professionals and engineers.
The team will develop a software tool to implement the automated analysis of crash narratives. This is particularly helpful to practitioners who manually read and analyze crash narratives when they need additional information or need to develop a more in-depth understanding than is available from reports. This can improve the work efficiency by 1) substantially reducing their time commitment (normally 10-15 minutes per one crash narrative by manual reading) and 2) improving outcome consistency and reliability. As also mentioned in the project description, the project will offer several other contributions as follows: 3) our improved understanding on crash factors can enhance decision-making at policy levels and aid in formulating effective strategies to enhance traffic safety, including targeted countermeasures and 4) this project supports recent endeavors in the transportation sector aligned with advancing transportation safety, if applied to other key parameters of interest by safety professionals and engineers.
PROJECT 2
FREEWAY AND ARTERIAL PERFORMANCE AND SAFETY ANALYSIS WITH HIGH RESOLUTION VEHICLE TRAJECTORY DATA
PROJECT DESCRIPTION
Local traffic agencies have large investments in intelligent transportation system (ITS) infrastructure including sensors such as cameras, radars, and loop detectors and communication to gain new insight for better planning, management, and operation of roadways. However, the ITS infrastructure is generally limited to dense urban areas and requires significant support to maintain coupled with limited in-house expertise to fully realize the promise of big data. Further, new high-resolution vehicle trajectory data (HRVT) streams have become available to further complicate analysis and the value proposition of ITS hardware. This project will evaluate the potential for HRVT to support infrastructure owner-operators (IOOs) and practitioners to more effectively plan, operate, and manage their systems and improve safety outcomes on their networks. HRVT will be integrated into traditional traffic analyses as well as leading-edge deep learning research. The outcome of this project will be a tool to effectively query, store, process, and visualize HRVT data for practitioner use.
PRINCIPAL INVESTIGATOR(s)
Brendan Morris
Shashi Nambisan
FUNDING SOURCE & AMOUNT
USDOT, $143,821 Federal, $52,843 Non-Federal
PROJECT START DATE
June 1, 2023
RESEARCH OUTPUT & IMPACTS
The successful completion of this project will offer several key values in transportation safety research and education. First, by replacing the time-consuming manual reading process with an automated approach based on NLP and AI, we will streamline operations and enhance efficiency in transportation safety analysis. Second, this initiative will significantly reduce (if not eliminate) the potential for errors and unreliable outcomes, ensuring the delivery of consistent and dependable results to analysts. Third, our improved understanding on crash factors can enhance decision-making at policy levels and aid in formulating effective strategies to enhance road safety, including targeted countermeasures. Last, this endeavor aligns with recent technological advancements and their practical implementation in advancing transportation safety, which can have a broader impact if applied to other key parameters of interest by safety professionals and engineers.
This project is expected to produce four major types of products: i) publications, ii) benchmark datasets, iii) open-source code/software, and iv) a prototype for a computer tool for HRVT data capture, integration, and analysis.
PROJECT 3
PROTECTING LOW INCOME ROAD USERS (COMMUNITY ENGAGEMENT, TECH TRANSFER)
PROJECT DESCRIPTION
The project will strive to mitigate crashes, with emphasis on outcomes that are fatal or life-altering with a focus on populations from lower socio-economic strata and neighborhoods. Such communities are especially likely to have lower access to automobiles, and greater reliance on public transportation or non-motorized modes of transport (i.e., for those who use human-powered transportation such as walking or bicycling). Thus, it will attempt to address the inequities in the transportation access and mobility needs of vulnerable population groups.
Research shows that in states like Nevada, where costs to own and operate a vehicle are high, families whose incomes are under $50K struggle to operate a vehicle, and those who earn under $25K are simply not able to. The working poor are those who are walking, biking, and taking transit to get to their jobs and are the ones most at risk for traffic incidents.
This project will utilize crash data, census data, and coroner data for residences and examine neighborhoods by crash history, income levels, and observed road behaviors. At identified high crash locations, we will utilize advanced sensor technologies (e.g. camera, radar, lidar), and autonomous technologies as well. The combination of both kinds of data will enable us to identify potentials for change and determine countermeasures required to address the identified problems. A comparative analysis will consider the issues on the street, the crash characteristics, and best practices to mitigate problems by household income levels and using the federal poverty guidelines.
We will use the electronic data we collect and the crash data for the most recent 5 years available, right now 2018-2022, and update city and county officials on the needs in their jurisdictions. We will work with neighborhood associations, elected officials as well as the planning and engineering departments of each entity to advocate for needed upgrades, potential projects, and outreach to the neighborhood. We will ensure that the road user is part of the discussion for identifying needs and developing countermeasures, in addition to law enforcement; two populations who are often not included in the discussion of issues that affect them every day.
PRINCIPAL INVESTIGATOR(s)
Erin Breen
Shashi Nambisan
FUNDING SOURCE & AMOUNT
USDOT, $31,228 Federal and $0 Non-Federal
PROJECT START DATE
June 1, 2023
RESEARCH OUTPUT & IMPACTS
The project efforts will be to collaborate with various stakeholders and infrastructure system owners to identify safety-related needs, challenges, and opportunities of populations from lower socio-economic strata and neighborhoods in the Las Vegas metropolitan area. This will involve numerous meetings and outreach sessions. It will compile data and configure by zip code to support Safe Streets for All (SS4A) initiatives in the region. It will provide an avenue to integrate the needs of such vulnerable road users (especially pedestrians and bicyclists) in the transportation access and mobility decision-making processes and practices.
The outcomes of the project will help inform decision-makers at the municipal, county, Metropolitan Planning Agency, and state levels to identify and prioritize critical needs to improve access and mobility needs of populations from lower socio-economic strata and neighborhoods. The outcomes could help develop related policies, design standards and guidelines, operating practices, and administrative processes as well as resource allocation matters.
PROJECT 4
EFFECTIVE PRACTICES TO INTEGRATE TRAFFIC CITATION AND ADJUCATION (TCA) DATA
PROJECT DESCRIPTION
This project aims to compile effective practices to integrate Traffic Citation and Adjudication (TCA) datasets and provide a synthesis of the same so as to enhance safety for road users. Further, it will outline critical steps to help adopt such practices in Nevada. Key project activities will include a review of the literature, federal, state, and local legislation and regulations, possible case studies, interviews with key individuals and organizations, and briefings of stakeholders.
Repeat violators of traffic laws pose a substantial risk and cause irreversible harm to road users. The disconnect between law enforcement officers (LEOs) / law enforcement agencies (LEAs) and the judiciary / judicial processes often leads to such individuals not being identified in a timely manner. Such identification would enable interventions that could avert avoidable adverse safety outcomes resulting from the actions of these road users. The disconnect may be due to policies, programs, practices, or processes related to completing the various steps between a LE officer interacting with a road user and any potential adjudication outcomes.
To try to avoid (or at least reduce the potential for) repeat violators of traffic rules and regulations causing irreparable harm to road users, key gaps in the “system” need to be eliminated. The gaps include procedural steps, timeliness of actions, and access to data such as gaps between when an LEO interacts with a motorist and issues a warning or a citation, how this is recorded in the database, how the data/records are made available with minimal latency to other LEOs, who has access to these databases, how these databases are linked to the court system, and how the adjudication process and their outcomes are linked back to the citation system.
PRINCIPAL INVESTIGATOR(s)
Shashi Nambisan
FUNDING SOURCE & AMOUNT
USDOT, $31,228 Federal and $0 Non-Federal
PROJECT START DATE
June 1, 2023
RESEARCH OUTPUT & IMPACTS
The project aims to analyze and synthesize effective approaches adopted by state and local agencies to integrate TCA data, lessons learned, and pertinent evaluations of the costs and benefits of the same. The project’s efforts will document effective approaches to integrate TCA data and to synthesize relevant legislative, regulatory, and voluntary approaches. It will focus on law enforcement agencies (LEAs) and Judicial Agencies (JAs) across Nevada who handle traffic citations, document the linkages (workflow paths?) between law enforcement officers (LEOs), LEAs, and JAs in Nevada, connect with key stakeholders, and initiate efforts to obtain relevant TCA data.
The outcomes of the project will help develop critically important policies, programs, procedures, and practices to integrate relevant but currently disparate data sets across state and local agencies in Nevada. These are vital to bridging the gap in the traffic safety records system so as to eliminate or reduce fatalities and injuries caused by repeat violators of motor vehicle in Nevada.
PROJECT 5
ADMINISTRATIVE TASKS
PROJECT DESCRIPTION
The administrative tasks are to coordinate various aspects of the grant. These include matters that are internal to UNLV and others that are external to UNLV such as with Howard University (the lead institution for REPS), other members of the REPS Consortium, engaging with advisory board members and stakeholders, complying with administrative and reporting requirements.
PRINCIPAL INVESTIGATOR(s)
Shashi Nambisan
FUNDING SOURCE & AMOUNT
USDOT, $65,181 Federal and $37,727 Non-Federal
PROJECT START DATE
June 1, 2023
RESEARCH OUTPUT & IMPACTS
The outputs of the administrative tasks of the grant will be the submittal of various administrative and management reports, tracking and reporting on the progress of individual projects within the grant at UNLV, and participating in meetings with advisory board members and stakeholders, as well as the US DOT representatives.
The outcomes of the administrative efforts will be effective monitoring and management of the grant’s activities; identifying relevant needs, challenges, and opportunities; and enabling informed decision-making for the allocation of resources in the future.