Source Sought Notice_693JJ3-26-SN-0010.pdf

SOURCES SOUGHT NOTICE 693JJ3-26-SN-0010

For

“AI-Enabled Incident Response System with sUAS and Human Copilots (AIRS-COP)”

1. SUBMITTAL INFORMATION

ISSUE DATE: September 30, 2026

DUE DATE FOR RESPONSES: October 15, 2026

SUBMIT RESPONSES TO: jason.motala@dot.gov

See below Section 4 entitled “REQUESTED CONTENT OF RESPONSES TO THIS SOURCES SOUGHT NOTICE” for submittal content instructions.

2. INTRODUCTION

This is NOT a solicitation for proposals, proposal abstracts, or quotations.

The purpose of this notice is to obtain information, for planning purposes, regarding: the availability and capability of qualified sources to perform the work, as well as the size and status of potential sources. The NAICS Code for this planned acquisition for “AI-Enabled Incident Response System with sUAS and Human Copilots (AIRS-COP)” is 541715 – Research and Development in the Physical, Engineering, and Life Sciences (except Nanotechnology and Biotechnology) – with a Small Business Size Standard of up to 1,000 employees.

All potential sources, regardless of whether they are large or small in accordance with the NAICS size standard are encouraged to submit a capabilities statement in response to this Sources Sought notice.

It is anticipated that the potential resulting solicitation will result in a single award labor hour with cost-reimbursable travel research and development award.

3. Statement of Work mailto:amy.morrissette@dot.gov

Statement of Work

For

AIRS‑COP: AI‑Enabled Incident Response System with sUAS and Human Copilots

1.0 Overview

This is a non-personal services purchase order to pilot an approach on AI‑Enabled Incident Response System with Small Unmanned Aircraft System (sUAS) and Human Copilots. Under this purchase order, the Contractor will develop AI technologies which extend beyond existing sUAS programs for traffic monitoring and data collection by leveraging AI to enable sUAS to collaborate with Traffic Management Centers (TMCs) and human responders to support rapid decision-making and provide autonomous first‑response traffic management prior to the arrival of human crews. The Government will not exercise any supervision or control over the purchase order service providers performing the services herein. Such service providers must be accountable solely to the Contractor who, in turn is responsible to the Government as defined in this Statement of Work (SOW). The Contractor must perform to the standards herein.

1.1 Background

Under 23 U.S. Code §503(b)(1), Highway Research and Development Program, the Secretary of the Department of Transportation is directed to carry out highway research and development that addresses current and emerging highway transportation needs. The objectives of the Highway Research and Development (HRD) Program are: (A) identify research topics; (B) coordinate research and development activities; (C) carry out research, testing, and evaluation activities; (D) provide technology transfer and technical assistance; (E) engage with public and private entities to spur advancement of emerging transformative innovations through accelerated market readiness; and (F) consult frequently with public and private entities on new transportation technologies.

The mission of the Federal Highway Administration (FHWA) is to improve safety and mobility on our nation’s highways through national leadership, innovation, and program delivery. As part of that mission, FHWA is the Operating Administration (OA) responsible to the Secretary to conduct research and provide technical assistance to state and local agencies to improve safety and mobility, and to encourage innovation, in accordance with the requirements of the HRD Program.

The Office of Safety and Operations Research and Development (HRSO) performs transportation safety and operations research and development (R&D) for the FHWA. Research that addresses crashes in which there is some combination of human factors, vehicle factors, physical environment factors, and socioeconomic factors; determining why traffic crashes occur, as well as, increasing the understanding of what leads to the severity of injury outcomes. HRSO conducts safety and operations R&D based upon a national perspective of the transportation needs of the United States (US).

The objective of this purchase order is to design, develop, and flight-test a smart copilot system including smart sUAS, TMC and law enforcement officers for highway incident first responder services. These smart sUAS and TMC operator copilot system will leverage innovative AI and sUAS technologies to reduce traffic incident response and traffic restoration time and improve traffic safety, which is authorized under 23 U.S.C. §503(b)(1)(C); and, 23 U.S.C. §503(b)(2)(C)(ii), (iii), (v) and (xvi), and 23 U.S.C. §503 (b)(5) C (x) (xvi) and (xx).

sUAS offers a transformative solution by providing a real-time, aerial perspective to assess highway incidents quickly and safely. The FHWA recognizes its great potential in enhancing safety and operations across the highway system. While the Federal Aviation Administration (FAA) handles regulations for sUAS flights, the FHWA works with state and local departments of transportation (DOTs) to develop and implement applications for ground transportation. For example, FHWA EDC-5 report claimed the sUAS can speed up data collection, increase safety and accuracy, and provide access to hard-to-reach locations when evaluating highway infrastructure. Alaska DOT and Public Facilities (DOT&PF) have begun deploying sUAS across the State to aid in design, construction and monitoring of large-scale projects.

NJDOT has received grants from FHWA to evaluate the use of sUAS for variety of purposes such as structural inspections. MassDOT has pursued FHWA grants to use sUAS and data analytics for highway inspections and maintenance. NCDOT has used sUAS for construction inspection to monitor and document complex projects, including work zone traffic monitoring. Both Texas DOTs and Delaware DOTs have started to use sUAS to assist data collection for traffic incident management. In short, existing projects focus investigating the potential benefits of sUAS in inspection and monitoring traffic which helps traffic safety and efficiency.

Based on recent engagement with state DOTs pioneering sUAS applications in Traffic Incident Management (TIM), there is a critical opportunity to pivot sUAS capabilities beyond current data collection to directly assist agencies and first responders in incident response and management.

The current state of the practice highlights significant challenges; delays in response often lead to prolonged congestion, an increased risk of secondary collisions, and slower medical aid.

Traditional methods—reliant on 911 calls, dispatchers, and human-driven patrols—are frequently limited by traffic conditions, geography, and line-of-sight constraints, resulting in lengthy incident reconstruction and recovery times.

This purchase order will develop a prototype, followed by future pilot implementation testing, of an AI-Enabled Incident Response System with sUAS and Human Copilots (AIRS-COP). This system integrates AI and sUAS technology into a sophisticated, autonomous, and human-supported framework to improve response efficiency. The application of sUAS in the AIRS-COP moves beyond passive data collection by utilizing "Agentic AI," acting as a smart agent that can automatedly and/or autonomously fly to the incident site and copilots ground traffic operators and law enforcement officers to conduct timely first response. The success of this project will accelerate the adoption of sUAS by state DOTs for faster, autonomous first-response actions that significantly improve safety and save lives. Supporting this initiative, academic research has provided a strong technical foundation through proof-of-concept studies on critical technologies.

For example, George Washington University has evaluated two state-of-the-art object detection models for aerial-view perspectives: the YOLOv8 (Convolutional Neural Network) and Real-Time Detection Transformer models (https://arc.aiaa.org/doi/10.2514/6.2025-3521). Their findings offer insights into the trade-offs between detection performance, inference speed, and uncertainty calibration, providing essential guidance for selecting models in safety-critical UAS landing applications. Additionally, their studies have investigated fine-tuning Large Language Models for cooperative tactical deconfliction of sUAS, yielding improvement for decision accuracy and consistency (https://arxiv.org/abs/2603.28561). This project will leverage rigorous academic research and practical operational experience to advance sUAS-enabled traffic incident response and management systems in state DOTs.

1.2 Objective

Briefly, the AIRS‑COP will extend beyond existing sUAS programs for traffic monitoring and data collection by leveraging AI to enable sUAS to collaborate with Traffic Management Centers (TMCs) and human operators therein as well as law enforcement officers to support rapid decision-making and provide autonomous first‑response traffic management prior to the arrival of needed human crews. Actions such as lane closures, route suggestions, and other traffic-control measures may be executed by the sUAS communicating with roadside traffic controller or directed by the TMC. This capability will be built on real‑time sUAS data collection and low‑latency transmission between sUAS edge compute and cloud resources. The system will explicitly partition responsibilities amongst AI-augmented sUAS, human operators/TM, and law enforcement officers to ensure safety and clear control authority. Correspondingly, the core objective of this research includes

• Design the AIRS‑COP architecture that integrates sUAS edge computing and cloud computing, accounts for battery and thermal constraints, and delivers recommended hardware/software trade‑offs and system specifications.

• Enable AI-aided sUAS to navigate autonomously or automatedly to approach incident sites based on real‑time observations and TMC instruction updates.

• Integrate vison large language models (VLMs) with the sUAS edge and cloud pipeline to support real‑time incident data collection, processing, first response, and communication to TMCs, human operators therein and law enforcement officers.

• Flight‑test the AIRS‑COP prototype (sUAS(s)/AI + TMC + human operators) to assess feasibility, core functionality, latency, and operational risks at a controlled and approved flight-site, such as the TFHRC campus or an equivalent test facility.

This project executes a full-scale development pipeline—from hardware/software proofs-of-concept development for AIRS-COP design to campus testing and pilot evaluation in state DOT. These outcomes will significantly accelerate the adoption of sUAS by state DOTs, streamlining traffic incident management and empowering faster, autonomous first-response actions.

1.3 Detailed Requirement

The Contractor must provide all personnel, sUAS, equipment, supplies, facilities, transportation, tools, tools, materials, supervision, and other items and non-personal services necessary to perform.

Task 1. Project Management

Task 1.1. Project Management Plan https://arc.aiaa.org/doi/10.2514/6.2025-3521 https://arxiv.org/abs/2603.28561

The Contractor must draft and submit a Project Management Plan (PMP) to the Technical Point of Contact (TPOC) 1 week before the kickoff meeting. The PMP shall be reviewed at the kick-off meeting. The PMP must include:

• Project schedule and work task dependencies.

• Staff labor/cost distribution by task.

• Communications management plan.

• Project risk management.

• Quality management plan.

• Propose FHWA Project Review Panel (PRP) candidates and their roles.

• Approach to reviewing draft reports / public affairs 508 compliance approval, and public affairs final acceptance approval.

Any subsequent changes to the PMP that affects the project schedule and labor hours need to be approved by the TPOC.

NOTE: The PMP shall be consistent with the order and all exhibits. The PMP shall not be construed to override or change the requirement or exhibits of this order. The controlling document for the requirement shall be the order, its exhibits, and any modifications issued.

Task 1.2. Kickoff Meeting The Contractor must schedule, create agenda, and attend a kickoff meeting with the FHWA purchase order TPOC within 2 weeks (10 business days) of the Order award.

The Contractor may be required to host the web conference and must be prepared to provide options. Key project personnel from the Contractor are expected to attend. The kick-off meeting with TPOC will

• Review the background and objectives of the project.

• Review roles, authorities, and responsibilities of individuals involved from the Contractor.

• Reach agreement on common issues, resolve, and identify potential problems, and clarify understandings of technical aspects of the purchase order, including specific details of the scope and its deliverables.

• Review FHWA PRP candidates and due dates and reach agreement with the TPOC.

• Review the approval process of the potential fight tests such as TFHRC campus or other sites, when is the best time to start this approval application process to match the schedule and scope of this effort.

• Review the availability of the data and computing resources for AI model development, and whether additional data or computing resources will be required.

• Identify refinements that need to be made to shape the final PMP.

• Describe risks and risk mitigation measures.

• Present, discuss and refine the project schedule as necessary after reviewing the scope and expected deliverable with TPOC.

• Receive further guidance from the TPOC and anything else needed to finalize the Project

Management Plan.

Following up with the kickoff meeting, the Contractor must develop a modified PMP. The modified PMP shall include all pertinent work information identified during the kickoff meeting. The Contractor must document minutes, actions, and decisions from the kick-off meeting as kick-off minutes.

The kick-off meeting summary and slides are due within 3 days following the kickoff meeting.

Finalize, invite, and confirm the members of FHWA Project Review Panel 2 weeks following the kickoff meeting. The Modified Final Project Management Plan is due within 2 weeks following the kick-off meeting.

Task 1.3. Bi-Weekly Project Meetings

Following the kick-off meeting, the Contractor must coordinate regular bi-weekly status meetings (not needed during the publication review process) with the TPOC and other FHWA staff. In a bi-weekly meeting, the Contractor must provide updates on the living document, progress, schedule, and risks. Regular status meeting agenda (notes) must be delivered by the Contractor to the TPOC no more than 2 business days before (after) the meeting. These status meetings must allow for the review team members to participate remotely via webinar format, preferably, and via telephone conference line at a minimum.

Task 1.4. Monthly Progress Review Meetings and Reports

FHWA anticipates monthly project review meetings as needed. Key project personnel shall attend, and the Contractor may be required to host the web conference and must be prepared to provide options. The agenda of the Monthly Project Review meetings may include:

• Summarize up-to-date project status, including but not limited to completed tasks, on-going tasks, budget and effort expenditure.

• Identify agreed-upon high-priority tasks for the next 4 weeks.

• Establish tangible acceptance criteria to ensure achievable goals in the next 4 weeks.

• Summarize encountered challenges and their resolutions.

• Document lessons learned.

• Rebaseline remaining tasks (scope, budget, or schedule updates), if needed.

The Contractor must submit Monthly Status Reports, which must include, but are not limited to, a purchase order project schedule, purchase order budget task by task, updates on work complete, hours and budget expended task by task, risk management log, and work forecasts.

Task 1 Deliverables:

• Draft and Final Project Management Plan

• Kickoff meeting notes and slides

• FHWA Project Review Panel

• Bi-weekly meeting notes

• Monthly project review meetings

• Monthly progress reports

Task 2. Literature Review, Stakeholder engagement, and Recommendation The Contractor must perform stakeholder engagement, a comprehensive review and synthesis of current technologies and notable best practices related to the use of sUAS for traffic incident management (TIM), using or not using AI.

Task 2.1 Literature Review and Recommendation The Contractor must conduct a comprehensive review of existing literature on small Unmanned Aircraft Systems (sUAS) applications in transportation, with a primary focus on safety and operational aspects, while broadly encompassing infrastructure use cases. This review will aim to identify state-of-the-art research developments that align closely with the project’s objectives, ensuring the findings are relevant and applicable by adhering to the following criteria:

1. Prioritize studies and projects demonstrating potential for real-world deployment, excluding purely theoretical work. Simulation studies may be considered only if they provide direct validation or clear applicability to practical implementations.

2. Emphasize technologies and methodologies that have been deployed or validated by State DOTs, NASA, and other relevant federal and industry sectors. Early-stage concepts should be included only if directly pertinent to the project’s scope and goals.

3. Focus primarily on literature published within the past five years, except where older works provide foundational insights or are critical to understanding technological adoption trends.

The review will incorporate the elements listed below.

1. sUAS Autonomous and Semi-Autonomous Navigation:

• Review studies and solutions enabling sUAS autonomous and semi-autonomous navigation specifically for traffic incident response scenarios.

• Assess navigation algorithms, path planning, obstacle avoidance, and control methodologies validated in real-world or field-test environments.

• Identify challenges, limitations, and emerging technologies supporting safe and reliable sUAS navigation in complex traffic settings.

• Recommend potential technical solutions for this effort.

2. Artificial Intelligence (AI) Integration, Especially Vision-Language Models (VLM):

• Review AI applications on sUAS focused on real-time data processing and automated decision-making relevant to traffic incident scenarios.

• Highlight approaches implementing LLMs or similar advanced AI models onboard sUAS or within related cloud/edge computing frameworks.

• Provide recommendations and insights for this effort.

3. Human-AI Collaboration and Computing Architectures:

• Examine current research on AI-enabled sUAS operating alongside human copilots or responders, emphasizing collaborative workflows and task-sharing.

• Analyze state-of-the-art approaches to distributing computation between sUAS edge devices and cloud infrastructure to optimize responsiveness, data management, and operational robustness.

• Analyzing the state of the art of the solution addressing computing, battery usage, and thermal management.

• Provide recommendations and insights for this effort.

4. Computing load distribution between sUAS edge and cloud

• Review the edge computing capacity of different types of sUAS

• Review existing schemes and best practices of sUAS edge and cloud computing load allocation

• Provide recommendations for this effort.

The findings will be summarized in a Technical Memorandum (Tech Memo) titled sUAS Research Gaps and Recommendations in Transportation Applications. The Memo will synthesize the state of art for the elements described above, documenting where and how these technologies are being implemented and tested, the issues or limitations encountered, and the relative advantages and disadvantages of each approach. It will also provide actionable insights and recommend technologies and strategies appropriate for this effort. The Contractor must prepare and present a briefing on the draft that summarizes key findings at the next bi‑weekly meeting (before or after submission of the draft). The briefing must include a slide deck up to 5 pages and will be allocated up to 10 minutes for presentation. The TPOC and PRP will review the draft of Technical Memorandum and provide comments to inform subsequent tasks.

Literature Review must be completed within the first 2 months after project award.

Task 2.1 Deliverable:

• Draft Tech Memo on sUAS Research Gaps and Recommendations

• Briefing on sUAS Application Best Practices and Gaps (up to 5 slides)

• Final Tech Memo on sUAS Research Gaps and Recommendations

Task 2.2 Stakeholder engagement

The objectives of this subtask are to engage internal and external stakeholders to evaluate the current state of practice for integrating AI into small Unmanned Aircraft Systems (sUAS) applications in transportation, with a primary focus on safety and practical applications, while also addressing infrastructure-related uses.

As part of this task, the Contractor must propose and identify a list of stakeholders including key state DOTs, such as those states within the Uncrewed Aircraft Systems (UAS) Standardization Transportation Pooled Fund study. The final stakeholder list is due 1 week after receiving FHWA feedback. Once the TPOC accepts the finalized stakeholder group, the Contractor must conduct stakeholder engagement activities to solicit internal and external input and share knowledge on sUAS Applications in Transportation. This includes but is not limited to:

• Summarize existing sUAS uses and implementations in transportation safety and operations with particular emphasis on traffic incident management (TIM).

• Characterize the types of sUAS platforms employed, their operational capabilities, and deployment contexts.

• Identify key research gaps, limitations, and recommendations for next steps to advance sUAS effectiveness in TIM.

• Levels of autonomy and human-in-the-loop procedures.

• Analyze strengths and weaknesses of different sUAS types relative to TIM needs and provide recommended sUAS type used in this effort.

• How does the standard practice in terms of the usage of sUAVs amongst TMC and law enforcement; what operational protocols are typically used.

Stakeholder engagement must be conducted through a hybrid approach, combining email-based information collection and virtual meetings. The Contractor must obtain agreement on the appropriate approach for each context. For email-based engagement, the Contractor will prepare targeted questions, submit them to the TPOC for review, and then distribute them to stakeholders. For webinars, the Contractor is responsible for preparing (e.g., developing slides, readahead materials, question lists, and agenda, securing teleconferencing services), organizing and facilitating the meetings (e.g., facilitating discussion, taking notes). The Contractor must comply with all requirements related to Public Affairs review of external facing materials when delivering the read ahead materials and presentation. The Contractor must deliver meeting notes from each stakeholder engagement meeting one week after the meeting(s) takes place. The number of stakeholder events related to developing the sUAS Application Best Practices and Gaps Report will be correlated with the size and structure of the proposed stakeholder engagement group, but FHWA anticipates 2 meetings and each meeting includes multiple stakeholders. No more than one of these two meetings is in-person. Stakeholder engagement events must be completed within the first 3 months after project award.

Insights gained from stakeholder engagement must be documented in an internal draft report, sUAS

Application Best Practices and Gaps, and presented to the TPOC and PRP to support informed decision-making for Tasks 3 through 6. To facilitate the review, the Contractor must prepare and present a briefing to the TPOC and PRP on the draft that summarizes key findings at the next bi‑weekly meeting (before or after submission of the draft). The briefing must include a slide deck up to 5 pages and will be allocated for up to 10 minutes for presentation.

Task 2.2 Deliverables:

• Draft stakeholder engagement list.

• Final stakeholder engagement list.

• Stakeholder engagement meeting materials (for each meeting held).

• Stakeholder engagement meeting minutes (for each meeting held).

• Draft sUAS Application Best Practices and Gaps.

• sUAS Application Best Practices and Gaps briefing.

• Final sUAS Application Best Practices and Gaps Report

Task 3: sUAS Automated and Autonomous Navigation

While traffic incidents are increasingly reported via 911, crowdsourcing (Waze/Google Maps), and Next- Generation 911 (NG911) multimedia, initial reports often suffer from spatial uncertainty. Inaccurate or "fuzzy" location data complicates the deployment of Unmanned Aerial Vehicles (sUAS) for rapid scene assessment. This task aims to develop and implement existing robust navigation algorithms that allow sUAS to transition from "reported coordinates" to the "exact scene" through a combination of predefined automated flight and intelligent autonomous search.

Subtask 3.1: Develop Automated Navigation Controller

Objectives: This subtask addresses the long-range phase of the navigation mission, focusing on efficiency and reliability. Namely, once receiving an incident report, this automated path-to-coordinate navigation enables the sUAS to reach the general vicinity of the incident as quickly as possible using existing infrastructure data.

Scope and Execution Approach Models/Algorithm Development. Technically, the Contractor must develop or customize existing automated navigation controller that can successfully ingest the "best available" coordinates from dispatch systems and calculate an optimized flight path while including the following capacities:

• Airspace Deconfliction: Integrating with unmanned traffic Management to avoid obstacles and restricted zones.

• Predefined Trajectory: Executing a high-speed "sprint" to the reported coordinates or estimated GPS waypoint and fly back once task is finished or battery inventor is not sufficient. The two-way path should be optimized considering the battery capacity and the estimated computing request at the incident site.

• Telemetric Handover: Maintaining a stable data link with TMC for real-time mission updates adapting to the error of initial report.

Models/Algorithm Development Execution Approach: To initiate approach development, the Contractor must develop an Automated Navigation Controller Development Plan (1~2 pages), subject to TPOC’s comments and approval before initiating development. During execution, the Contractor must maintain and develop the plan as a technical living document for models and/or algorithms development, updated 2 days prior to each bi-weekly meeting to reflect current progress and updates. During these meetings, the

Contractor will use the technical living document to brief the TPOC on development status, collect comments, and address emerging technical issues. This technical living document is mainly used to track progress, exchange ideas, and communicate development. The purchase order must not spend many editorial hours on preparation. It will serve as the foundation for the Draft and Final Technical Memoranda of the models/algorithm Development in this task/subtask.

Safety and Efficiency Test and validation: The developed algorithm should be initially tested and validated in simulation to ensure the major functionalities and then fully tested in field (such as TFHRC campus and other approved flight sites) to validate the practical applications. Simulation and field tests should provide insights, including but not limited to the questions below:

• How quickly can the sUAS reach the mission within what range.

• How efficiently can the sUAS avoid obstacles and restricted zones under diverse environmental and operational scenarios.

Testing must be conducted across diverse environmental and operational scenarios such as windy, foggy, dark, rain, snow, and locations with barriers, like tree lines or urban canyons as well as other relevant conditions. The Contractor will plan the scenario list and collaborate with the PRP and obtain TPOC approval on the test site and scenarios selections.

Test Development Execution Approach: To prepare for the test, the Contractor must develop a Test Design Plan (including site and scenario list and testing setting), requiring TPOC comments and approval of the Final Test Design Plan prior to test commencement. During execution, the Contractor must maintain and develop the plan as a Living Test Development Document, updated 2 days before each bi-weekly meeting to reflect real-time progress and adjustments. In the bi-weekly meetings, the Contractor will use the living document as a base to brief the TPOC on testing progress, collect comments, and resolve any emerging issues identified during execution. This technical living document is mainly used to track and communicate the ideas, progress, and development. The contractor must not spend many editorial hours on preparation.

Upon test completion, the Contractor must develop the living documents of the task/subtask to deliver Draft and Final Tech Memo integrating all controller (model and algorithm) development and testing efforts.

These documents provide a comprehensive account of the model and algorithm development, codebase and scripting architecture, testing execution, and identified lessons learned.

Task 3.1 Deliverable:

• Draft Plan for automated navigation models and/or algorithms development (1~2 pages)

• Final Plan for automated navigation models and/or algorithms development (1~2 pages)

• Living documents for automated navigation models and/or algorithms development

• Draft Tech Memo for automated navigation models and/or algorithms development

• Final Tech Memo for automated navigation models and/or algorithms development

• Draft Plan of test design (1~2 pages)

• Final Plan of test design (1~2 pages)

• Living Document for test development.

• Draft and Final Tech Memo for integrated model and algorithm development, coding

(scripts and codebase structure), testing, and lessons learned

Subtask 3.2: Develop Autonomous Localization and "Last-Mile" Search

Objectives: This subtask addresses the short-range phase where the sUAS must compensate for inaccurate reporting, focusing on intelligence and computer vision. Namely, if the incident is not at the exact reported GPS coordinate, the sUAS must "find" it.

Scope and Execution Approach Models/Algorithm Development. Upon arriving at the initial waypoint, the sUAS will transition from automated flight to an autonomous search pattern. The Contractor must develop or customize an AI-based autonomous navigation algorithm (e.g., Visual SLAM (Simultaneous Localization and Mapping)) for scene discovery that successfully enables the functionalities, including but not limited to:

• Computer Vision: Using onboard sensors, such as cameras or LiDAR, to detect visual cues of an incident (e.g., flashing lights, stopped vehicles, debris).

• Dynamic Path Planning: Executing an onboard AI algorithm (such as deep learning) that will adjust the flight path in real-time based on visual confidence scores, effectively "homing in" on the true incident site. The AI algorithm should be optimized to save battery usage while satisfying computing requests.

• Obstacle Avoidance: Utilizing on-board sensor data (such as LiDAR or stereovision) to navigate complex environments (e.g., overpasses, trees, or power lines) without human intervention.

Safety and Efficiency Test and validation: The developed algorithm must be validated in simulation to ensure the major functionalities and then fully tested in field (such as TFHRC campus and other approved flight sites) to validate the practical applications. Simulation and field test should provide insights, including but not limited to the following aspects:

• How sUAS can smoothly switch from automated navigation to autonomous navigation.

• How fast sUAS can accurately locate and approach the mission.

• What are the computing requirements, and could the sUAS accommodate the computing reasonably?

• What is the battery usage and thermal performance.

Testing must be conducted across diverse environmental and operational scenarios—including windy, foggy, low-light (dark) conditions, rain, snow, and locations with physical barriers such as tree lines or urban canyons—as well as other relevant conditions (e.g., temperature extremes, and GPS‑challenged environments) to ensure robust evaluation. The Contractor will plan the scenario list, collaborate with the PRP, and obtain TPOC approval on the test sites and scenarios selections.

Execution: The Contractor must execute Task 3.2 following the same development and deliverable reporting procedures established in Task 3.1.

Task 3.2 Deliverable:

• Draft Plan for autonomous navigation models/algorithms development (1~2 pages)

• Final Plan for autonomous navigation models/algorithms development (1~2 pages)

• Living documents for autonomous navigation models and/or algorithms development

• Draft Tech Memo for autonomous navigation models and/or algorithms development

• Final Tech Memo for autonomous navigation models and/or algorithms development

• Draft Plan of test design (1~2 pages)

• Final Plan of test design (1~2 pages)

• Living Document for test development

• Draft and Final Tech Memos for integrated model and algorithm developing, integration of Task 3.1 and Task 3.2, coding (scripts and codebase structure), testing, and lessons learned.

Task 4: AI-Augmented sUAS as the First Responders

Objectives: This task focuses on transforming sUAS into proactive AI agents (AIRS‑COP) by integrating Edge/Cloud computing with advanced Vision-Language Models (VLM). The objective is to transition from passive video streaming to active scene intelligence, enabling the sUAS to interpret complex traffic incidents and communicate actionable insights to TMC and the operators.

Scope and Execution Approach:

Models/Algorithm Development: The contactor must develop a VLM-driven pipeline that triggers upon arrival at an incident site and enable the sUAS as the first responder to successfully conduct intelligent scene perception and real-time synthesis, including but limited to

• Uses VLMs to make the AI agent translate visual data into immediate traffic operation recommendations associated with traffic incident types/scenes to TMC. The Contractor must propose potential and practical recommendations (e.g., suggest specific ramp closures or emergency lane deployments) for the PRP for discussion and TPOC for approval.

• Uses VLMs to enable contextual understanding and move sUAS’s current usages beyond simple object detection (e.g., "car") to complex scene reasoning (e.g., "multi-vehicle collision with fuel leak blocking two lanes").

• Select representative scenes and datasets to train the capabilities of the AI-Augmented sUAS. The Contractor must propose scenes and datasets by consulting PRP and for approval from TPOC.

• Uses LLM to make the AI agent understand the language/text instruction from operators at TMC or law enforcement officers and other human responders.

To develop this AI-agent, the Contractor should also analyze the computational "load-balancing" required for real-time autonomy, and optimize the hybrid computation, including but limited to

• Evaluate the feasibility of running compressed VLMs on sUAS’s hardware for low-latency decision-making and privacy-sensitive data processing.

• Identify the best practices of offboard cloud/TMC integration with edge computing in this sUAS application, factoring computing request, battery limitation and the thermal issues.

• Assess the network resilience and understand how the AI agent performs during intermittent connectivity when switching between edge and cloud resources under different operation environments.

Task 4 will provide the options for Task 5 to select the responsibility/functionalities of sUAVs under different incident scenarios.

Safety and Efficiency Test: The safety and efficiency of the AIRS‑COP (an AI-agent) must be tested in the simulation and field environment (such as at TFHRC campus leveraging the crash test the safety group in HRSO), and provide insights including but not limited to

• Verifying the generated texts and scene understanding from the VLM is trustworthy.

• How efficiently the hybrid computation? For example, how long the sUAS needs to fly back and be recharged under what scenarios

• How efficiently AIRS‑COP can conduct contextual understanding and traffic control recommendations? For example, please specify what traffic scenarios will function well while others do not. Quantify the improvement in traffic restoration and corresponding traffic benefits.

The test should be done under diverse environmental and operational scenarios as identified in Task 3. The Contractor must collaborate with the PRP and obtain TPOC approval on the test sites and scenarios selections.

Execution: The Contractor must execute Task 4 following the same development and deliverable reporting procedures established in Task 3.1.

Task 4 Deliverables:

• Draft Plan for models and/or algorithms development (1~2 pages)

• Final Plan for models and/or algorithms development (1~2 pages)

• Living Document for models and/or algorithms development

• Draft Tech Memo for models and/or algorithms development

• Final Tech Memo for models and/or algorithms development

• Draft Test design plan (1~2 pages)

• Final Test design plan (1~2 pages)

• Living Document for test development

• Draft and Final Tech Memos: Detailed documentation covering design and deployment of AI-Agent for sUAS First Responders, algorithm development, codebase architecture, experimental results, and a "Lessons Learned" analysis for future scalability.

Task 5 sUAS-TMC CoPilot

Objectives: Bulit upon the findings in Task 2.2 for the standard practice and protocols in terms of the usage of sUAVs amongst TMC and law enforcement officers, this task investigates the collaborative synergy of the AIRS-COP and explores the collaborative interactions among AI‑augmented sUAS, Traffic Management Center (TMC) and its human operators, and law enforcement officers.

Scope and Execution:

Model/Algorithm/Test development: The Contractor must perform the activities, including but not limited to the following list:

• Develop rigorous Human‑in‑the‑Loop (HITL) and Hardware‑in‑the‑Loop (HWIL) framework and experiments that leverage the AIRS‑COP framework produced in Tasks 3 and 4. Experimental testing must be performed at TFHRC, utilizing the HRSO Safety Group’s vehicle crash‑test resources or other approved flight‑test facilities as appropriate.

The testing must provide comprehensive coverage of representative traffic incident scenarios, and physical environment.

• Built upon the HITL and HWIL experimental platform, the Contractor must develop and optimize the strategies for both (i) computational load balancing between edge and cloud assets and (ii) for cognitive (decision‑making) load sharing between the AI agent and TMC operators to maximize performance, responsiveness, and operator situational awareness.

This may include the studies including but not limited to the following list:

1. Define roles and responsibilities of TMC and its operators. For example, specify the responsibilities of the TMC and individual operators for dispatching, supervising, and managing AI‑augmented sUAS.

2. Define sUAS on‑scene responsibilities. For example, specify the actions, data collection tasks, and constraints for sUAS upon arrival at incident sites.

3. Defines the specific operational roles and responsibilities of law enforcement officers—from the initial dispatch phase through on-site arrival—ensuring seamless coordination with automated systems and traffic management centers.

4. Identify decision‑support information thresholds. For example, establish minimum data and information requirements (e.g., video resolution, descriptive text) necessary for sUAS and operators to make confident response decisions, including authorization to dispatch secondary resources (e.g., law enforcement, EMS, towing).

5. Establish communication protocols. For example, specify the communication channels, message formats, timing requirements, and fallback procedures for interactions among sUAS, AI agents, TMC systems, and human operators.

6. Establish decision‑making hierarchies and coordination schemes. For example, specify authority levels, collaboration and coordination mechanisms for AI‑assisted decisions, operator interventions, and law enforcement officers’ coordination during incident response.

Execution: The Contractor must execute Task 5 following the same development and deliverable reporting procedures established in Task 3.1.

Task 5 Deliverables:

• Draft Plan of the HITL and HWIL framework development (1~2 pages)

• Final Plan of the HITL and HWIL framework development (1~2 pages)

• Living Documents for HITL and HWIL framework development

• Draft Plan of computation and cognitive balance optimization (1~2 pages)

• Final Plan of computation and cognitive balance optimization (1~2 pages)

• Living Documents for computation and cognitive balance optimization

• Draft and Final Tech Memos: Detailed documentation covering design and deployment of the HITL and HWIL testing platform and experiments, algorithm development, codebase architecture, experimental results, and a "Lessons Learned" analysis for future scalability.

Task 6: Documentation and Conference Presentation

The objective of this task is to document the developed the AIRS‑COP and conduct a conference presentation or prepare conference and journal papers to introduce the AIRS‑COP to the community.

• The Contractor must produce integrated and detailed project report covering the AIRS- COP’s design, development, functionalities, user instructions, testing, and technical specifications and lesson learned for future scalability. The report must build upon the technical Tech Memos developed in Tasks 2-5.

• The Contractor must draft the conference and/or journal papers to introduce the technical development and testing approaches and results to the community.

• The Contractor must prepare a presentation deck for the AIRS‑COP to be delivered at the designated stakeholder meetings.

• The Contractor must present at a conference agreed by the TPOC.

• The Contractor must complete all editorial requirements for project report, slide deck, conference and journal drafts, including Section 508 compliance before HCC and HPA approval.

Task 6 Deliverables:

• Draft Presentation for the AIRS‑COP development

• Final Presentation for the AIRS‑COP development

• Draft Conference Papers

• Final Conference Papers

• Draft Journal Papers

• Final Journal Papers

2.0 Period of Performance

The period of performance will be 18 months from effective date of award.

3.0 General

3.1 Hours of Operation

Under this purchase order, the Contractor is responsible for conducting business, between the hours of 8:00 am and 6:00 pm Monday thru Friday except Federal holidays or when the Government facility is closed due to local or national emergencies, administrative closings, or similar Government directed facility closings. When hiring personnel, the Contractor must keep in mind that the stability and continuity of the workforce are essential.

3.2 Place of Performance

The work to be performed under this purchase order will be performed at the Turner-Fairbank Highway Research Center, Contractor/subcontractor facilities, and at offsite test facilities, as appropriate.

3.3 Physical Security

The Contractor must be responsible for safeguarding all government equipment, information, and property provided for Contractor use. At the close of each work period, government facilities, equipment, and materials must be secured.

3.4 Special Qualifications

The Contractor should own certified sUAS that are selected and to be used in this purchase order.

The Contractor should also possess and maintain the approval from FAA for flying sUAS in approved sites during the execution.

The Contractor certifies that, to the best of its knowledge and belief, the Unmanned Aircraft System (UAS), components, firmware, software, and associated services proposed for performance under this Contract do not include any covered UAS manufactured or assembled by a covered foreign entity as defined in Section 1825 of the FY24 NDAA and implementing regulations. The Contractor must:

(a) Provide, upon request, full disclosure of manufacturers, assembly locations, supply chain sources, firmware/software provenance, and any subcontractors involved in the supply chain for the UAS and associated systems.

(b) Immediately notify the Contracting Officer and the TPOC if the Contractor becomes aware of any condition that would cause the Contractor’s certification to be inaccurate, including component substitution, change in manufacturing location, or acquisition of supplier assets.

(c) Provide, upon request, supporting traceability documentation sufficient to validate the origin and chain of custody of UAS hardware and software components.

(d) Allow FHWA or their authorized agents access to supplier documentation and facilities for inspection or audit, subject to protection of proprietary information as provided under the purchase order.

(e) Agree that any material misrepresentation in the Contractor’s certification constitutes a material breach and may result in purchase order termination, recoupment of funds, and other remedies.

The Contractor must implement and maintain measures necessary to protect Federal information processed, stored, or transmitted by the UAS and associated systems consistent with OMB M-26-02. Such measures must include, at a minimum:

(a) Authentication and access control for ground control stations consistent with NIST SP 800-63 guidance, including multifactor authentication where remote access is permitted.

(b) Firmware and software updates provided only through trusted sources; testing of integrity prior to field deployment; use of isolated systems for update procedures.

(c) Encryption of Federal mission data in transit and at rest using validated cryptographic modules and algorithms approved by FHWA.

(d) The ability for FHWA to opt out of any forced data uploads or remote storage, and the ability to control the destination of Federal data.

(e) Support for remote security controls as appropriate (remote wipe/lock) and documented procedures for invoking remote controls.

The Contractor must maintain commercial general liability and aviation liability insurance that covers UAS operations. The Contractor must ensure that any PIC performing under this purchase order holds a current FAA Remote Pilot Certificate and has completed at least five flights on the specific UAS platform proposed for operations, unless otherwise approved by the TPOC.

All UAS operations must include a flight plan, which at a minimum includes: PIC, airspace review, standard weather briefing, area to be flown, limitations, obstacle clearance, purpose of flight, time of flight, expected duration of flight, communication plan, and emergency/contingency procedures. The use of a Visual Observer is required. All documentation must be provided to the TPOC at least 7 days in advance of the anticipated UAS operation date.

3.5 Periodic Progress Meetings

The Contractor agrees to attend progress meetings. The Contracting Officer or TPOC, and other Government personnel, as appropriate, may meet periodically with the Contractor to review the Contractor's performance. At these meetings, the CO will appraise the Contractor of how the government views the Contractor's performance and the Contractor will appraise the Government of problems, if any, being experienced. Appropriate action must be taken to resolve outstanding issues. These meetings must be at no additional cost to the government.

3.6 Identification of Contractor Employees

All contractor personnel attending meetings, answering Government telephones, and working in other situations where their contractor status is not obvious to third parties are required to identify themselves as such to avoid creating an impression in the minds of members of the public that they are Government officials. They must also ensure that all documents or reports produced by Contractors are suitably marked as Contractor products or that Contractor participation is appropriately disclosed.

3.7 Contractor Travel

The Contractor must be required to travel throughout CONUS during the performance of this purchase order to attend meetings and conferences. The Contractor is required to travel to the TFHRC and designated local sites for field flight testing. While on-site, the Contractor must adhere to all applicable facility policies and safety protocols. All travel requires Government approval/authorization and notification to the TPOC.

The travel budget for the entire period of performance is $2,000.This amount cannot be exceeded.

3.8 Materials

The Procurement of materials, including sensors and cloud infrastructure, may be required for the performance of this purchase order. All such purchases must be coordinated with DOT in advance. Some published material like research papers or articles could be procured, if deemed relevant and necessary. All procurement requires Government approval/authorization and notification to the TPOC.

3.9 Organizational Conflict of Interest

Contractor and subcontract personnel performing work under this award may receive, have access to, or participate in the development of proprietary or source selection information (e.g., cost or pricing information, budget information or analyses, specifications or work statements, etc.), or perform evaluation services which may create a current or subsequent Organizational Conflict of Interests (OCI) as defined…