693JJ3-24-SS-0008.pdf

SOURCES SOUGHT NOTICE

For Federal Highway Administration (FHWA)’s

Office of Safety and Operations R&D

SOURCES SOUGTH NUMBER: 693JJ3-24-SS-0008

“Development of Pedestrian Realistic Artificial Datasets (RADs)”

THIS IS A SOURCES SOUGHT NOTICE; BEING RELEASED PURSUANT TO

FEDERAL ACQUISITION REGULATION (FAR) PART 10: MARKET RESEARCH.

This SOURCES SOUGHT NOTICE is issued solely for informational, market research, and planning purposes only and does not commit the Government to any further contract actions (i.e., solicitation and/ or contract). Respondents are advised that the United States Government will not pay for any information or administrative costs incurred in responding to this SOURCES SOUGHT NOTICE. Not responding to this SOURCES SOUGHT NOTICE does not preclude participation in any future Request for Proposals/Quotations (RFP/RFQ) that may be issued. Responses to this notice are not offers and will not be accepted by the U.S Government to form a binding contract.

The U.S. Department of Transportation (DOT) / Federal Highway Administration (FHWA) is conducting market research to determine if there are any small business, academic institutions etc., capable of developing Pedestrian counts (volume/exposure) Realistic Artificial Dataset(s) (RADs) for the Office of Safety & Operations Research & Development located at the Turner-Fairbank Highway Research Center in McLean, VA.

An overview of the required services, objectives is provided below.

1. Introduction

The Federal Highway Administration (FHWA) is a part of United States Department of Transportation (USDOT). The FHWA Office of Safety and Operations Research and Development is dedicated to advancing the safety and efficiency of the nation’s highway transportation system through innovative research and development. Its mission is to develop and implement innovative solutions and advanced technologies to enhance the safety, efficiency, and reliability of the U.S. highway system.

Purpose of Sources Sought Notice

Office of Safety and Operations R&D

Through this notice, the FHWA’s Office of Safety and Operations R&D seeks input from academia, government, business, and industry groups on developing Realistic Artificial Dataset(s) (RADs) for Pedestrian Volumes. A RAD is a synthetic data set that is generated for which all the relationships between the variables are known and is supposed to reflect real world data.

A draft statement of work SOW has been developed with intention of accomplishing the project in two phases. The first phase will a) investigate the availability of pedestrian data and b) determine the data gaps for pedestrian safety research and developing crash modification factors for pedestrian crashes. The second phase will develop the RADs to fill some of the identified data gaps in phase one, with an emphasis on generating pedestrian volumes. We also welcome input on the SOW and any ideas on accomplishing the project.

2. Project Description

Background and Scope:

Historically, crashes have been viewed as the result of an error on the part of the individual roadway user—a law or rule was broken, a wrong decision made. Humans make mistakes while driving and, importantly, are vulnerable to the forces that occur during a crash. By focusing on eliminating fatal and serious injuries the inherent priority is placed on pedestrians and bicyclists, who are at a higher risk of fatal or serious injury than a person driving or traveling in a motor vehicle. Additionally, FHWA submitted a Complete Streets Report to Congress (CSRTC) (https://highways.dot.gov/sites/fhwa.dot.gov/files/2022- 03/Complete%20Streets%20Report%20to%20Congress.pdf) that outlined goals that support safe and comfortable transportation systems for all users and all abilities. Thus, understanding the behavior of pedestrian and bicyclist activity and their interactions with various roadway and roadside elements is key to supporting this goal of eliminating serious and fatal injuries. For example, pedestrian and bicycle exposure data can help define non-motorized roadway activity and support complementary analyses (e.g., safety, operations, planning); however, collection of this data continues to be problematic for both researchers and practitioners, as it is hard to obtain data that is both accurate and robust.

https://highways.dot.gov/sites/fhwa.dot.gov/files/2022-03/Complete%20Streets%20Report%20to%20Congress.pdf https://highways.dot.gov/sites/fhwa.dot.gov/files/2022-03/Complete%20Streets%20Report%20to%20Congress.pdf

Office of Safety and Operations R&D

Determining pedestrian counts is critical for meaningful pedestrian safety research and for developing countermeasures and crash modification factors (CMFs). Historically, collecting pedestrian volumes is done by manually counting the pedestrians at a location for a period of time and applying a model to get the counts for a longer period of time like Average Annual Daily Traffic (AADT). This method is resource intensive and not practical in large scale application. Additionally, researchers and practitioners have explored other methods for pedestrian data collection including automated methods (e.g., infrared sensors, cameras, etc.) and tangential counting (e.g., GPS data, Smartphone data, crowd source data, etc.); however, both of these methods still have limitations related to equipment dependency (e.g., feasibility, cost, and capability of equipment) and user activity compliance (e.g., pedestrian has a phone and app with them at all times), respectively. To help bridge the gap, modeling methods have been used to estimate pedestrian volumes including predictive statistical models, machine learning, and most recently realistic artificial data. Both statistical models and machine learning rely heavily on input data, and often are dependent on computational resources and limited by user assumptions. All data collection methods have their advantages and disadvantages; however, a hybrid approach could be used to blend multiple methods to improve prediction and accuracy of pedestrian volumes. One possible hybrid solution is to generate realistic artificial pedestrian volumes by considering interconnected attributes that affect pedestrian counts and developing these counts for various roadway functional classifications.

RAD can be a useful tool for research on highway safety. Specifically, areas where the use of RAD could be beneficial are determining a sample size, assessing a new estimation method, comparing methods to analyze alternatives, evaluating ways to generate multivariate models, and/or filling in data gaps such as pedestrian volumes for Vulnerable Road User safety research. Developing safety models and CMFs for pedestrian safety requires identifying causal relationships in the data, the use of average values for variables, variable errors, missing variables and complex variable dependencies, and the use of simple mathematical functions. RAD offers an innovative way to address these challenges by developing models that can accurately predict crash frequencies, explain the cause-and-effect relationships between pedestrian crashes and independent variables, and be used to develop models that generate pedestrian volumes which presents a major gap in developing any CMF.

The lack of pedestrian exposure for non-traditional functional classifications such as: Rural, Rural Town, Suburban, Urban and Urban Core is critical gaps in performing any meaningful pedestrian research and developing pedestrian safety countermeasures.

The RAD or RAD-generating software can be developed by gathering data in a variety of ways such as, but not limited to:

1. Knowledge of subject matter experts (SME),

2. Body of research performed, and statistical models developed on the given subject,

3. Formulas present in literature and design guides such as AASHTO’s

A Policy on Geometric Design of Highways and Streets (Greenbook) (https://aashtojournal.transportation.org/aashto-releases-7th-edition-of-its-highway-street-design-green-book/)

4. Simulating data from real world scenarios, and

5. Employing Artificial Intelligence and Machine Learning techniques to generate data.

Under FHWA’s Exploratory Advanced Research (EAR) program, two software programs that generate RADs have been developed, MIMIC and DREDGE. MIMIC was developed by University of Missouri team and DREDGE was developed by University of Connecticut and University of Central Florida teams. Following are the links to the final reports of the two projects and general description of RADs:

https://www.fhwa.dot.gov/publications/research/ear/20047/

MIMIC https://highways.dot.gov/sites/fhwa.dot.gov/files/FHWA-HRT-23- 015.pdf

DREDGE Vol I https://highways.dot.gov/sites/fhwa.dot.gov/files/FHWA_HRT_23_121.pdf

DREDGE Vol II https://highways.dot.gov/sites/fhwa.dot.gov/files/FHWA- HRT-23-122.pdf https://aashtojournal.transportation.org/aashto-releases-7th-edition-of-its-highway-street-design-green-book/ https://aashtojournal.transportation.org/aashto-releases-7th-edition-of-its-highway-street-design-green-book/ https://www.fhwa.dot.gov/publications/research/ear/20047/ https://highways.dot.gov/sites/fhwa.dot.gov/files/FHWA-HRT-23-015.pdf https://highways.dot.gov/sites/fhwa.dot.gov/files/FHWA-HRT-23-015.pdf https://highways.dot.gov/sites/fhwa.dot.gov/files/FHWA_HRT_23_121.pdf https://highways.dot.gov/sites/fhwa.dot.gov/files/FHWA-HRT-23-122.pdf https://highways.dot.gov/sites/fhwa.dot.gov/files/FHWA-HRT-23-122.pdf

Office of Safety and Operations R&D

These RAD-generating codes were developed to test the study design and analysis methods, rank the models developed by the researchers and determine if different study designs and analysis approaches are effective in development of the models. However, the RADs to be developed for the current project will have a different purpose. The artificial pedestrian volumes generated would be used as an input (exposure) to any potential pedestrian safety research, planning, design, operations of pedestrian facility and evaluation of any Complete Streets application for pedestrians in the absence of actual pedestrian counts.

3. Specific Information Requested

Please provide answers to following questions:

• Have you developed any synthetic or realistic artificial datasets? How were they used? How were they validated and what level of fidelity was achieved?

• Can you characterize the computational requirements you needed to generate the realistic artificial datasets you developed?

• What sources of input data were used to develop the RAD and what were the outcomes?

• Have you developed any pedestrian realistic artificial datasets and what was the purpose i. e. research, planning, operations etc.?

• If you were to develop a pedestrian realistic artificial dataset for pedestrian exposure, what inputs would you include to generate the most accurate data and why?

• How would you ensure or include flexibility in the dataset provided?

• How would you ensure quality and diversification in the input data to generate the RAD?

• Once the RAD(s) is/are developed, how would you check quality of the data making sure that values are within the realistic range?

• How would one validate the final pedestrian realistic artificial data sets?

• Once the RAD(s) is/are developed, how would you promote/market the product so that it is used by a wide range of stakeholders as a tool for research, planning and operation of pedestrian facility?

Office of Safety and Operations R&D

4. Instructions for the SOURCES SOUGHT NOTICE Responses:

Responses shall be submitted electronically to the e-mail addresses below:

EMAIL: bob.p.brown@dot.gov HCFA-22sri@dot.gov

Subject Line: 693JJ3-24-SS-0008 - Sources Sought Notice Response

ATTN: Robert P. Brown

Responses shall be submitted on or before October 11, 2024; 1:00 pm EST

- Format and Content:

• White paper/thesis format on 8 ½ x 11 inch paper. Pages that exceed this size, such as foldouts, are not allowed.

• The document can be either a PDF or Word format.

• Narrative text must be printed using a font size no less than 12-point font.

• Tables are permitted. Text in tables may be smaller than 12-point but must be legible.

• Page margins must be a minimum of 1-inch top, bottom, and each side.

• Page numbers may be located within 1-inch margin.

• Page Limits: The number of pages should not be more than 10 pages single sided. The page limit includes all narratives, figures, tables, appendices and all other ancillary materials with the exception of the following.

• Exceptions on Page Limits: The following items do NOT count against the page limitation:

o Document covers;

o Cover letters;

o Title pages;

o Divider pages;

o Tables of contents;

o List of acronyms;

mailto:bob.p.brown@dot.gov mailto:HCFA-22sri@dot.gov

• Proprietary information submitted in response to this SOURCES SOUGHT NOTICE will be protected from unauthorized disclosure as required by the Federal Acquisition Regulation (FAR). All proprietary markings should be clearly delineated. The respondent shall identify where data is restricted by proprietary or other rights and mark it accordingly.

The cover page shall contain:

• Company name

• Primary Point of Contact

• Phone Number and Email Address

• DUNS/Cage Code

• NAICS Code(s)

• Business Size

• Socio Economic Category (i.e., HBCU, Women Owned Small

Business, Veteran Owned Small Business, etc.)

• Federal Supply Schedule (FSS) or OASIS Contract Number and SIN, if applicable.

Technical Ability: Respondents shall submit a capabilities statement package to include an overall description of the proposed project and/or any product solution(s) and provide technical data and a demonstrated ability for those areas identified. Please ensure the following information is included:

1. The respondents should provide their relevant experience in developing similar projects, examples of previous work, qualifications, past projects, and client references for completing the project.

2. Resources available such as corporate management and key personnel to be assigned to tasks under this effort to include professional qualifications, and specific experience of such personnel.

3. Statement regarding capability to obtain the Department of

Transportation (DOT) security clearances for personnel.

4. Please note that under a Small-Business Set-Aside, in accordance with FAR 52.219-14, the small business prime must perform at least 50% of the work themselves in terms of the cost.

5. Company's ability to begin performance upon contract award.

Note: Responses may include links or web addresses that demonstrate evidence of your technical expertise (i.e.: final reports, technical Journals, peer reviewed papers, etc.).

4.1 SOURCES SOUGHT NOTICE Response Contact

Respondents to this notice shall designate a primary and one alternate point of contact within the company (Name, Title/Position, Address, Email, and Telephone).

4.2 Clarification of SOURCES SOUGHT NOTICE Responses

FHWA reserves the right to seek additional information from respondents identified with unique solutions that are determined to be beneficial to FHWA. To fully comprehend the information contained within a response to this notice, there may be a need to seek further clarification from respondents. This clarification may be requested in the form of brief verbal communication by telephone; written, electronic communication; or a request for a presentation of the response to a specific FHWA group or groups.

5. Next Steps:

The responses will be reviewed within 60 days of receipt. A subsequent Request for Proposals (RFP) will be issued after completion of the review.