<rss version="2.0" xmlns:atom="https://www.w3.org/2005/Atom">
  <channel>
    <title>Research in Progress (RIP)</title>
    <link>https://rip.trb.org/</link>
    <atom:link href="https://rip.trb.org/Record/RSS?s=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" rel="self" type="application/rss+xml" />
    <description></description>
    <language>en-us</language>
    <copyright>Copyright © 2026. National Academy of Sciences. All rights reserved.</copyright>
    <docs>http://blogs.law.harvard.edu/tech/rss</docs>
    <managingEditor>tris-trb@nas.edu (Bill McLeod)</managingEditor>
    <webMaster>tris-trb@nas.edu (Bill McLeod)</webMaster>
    <image>
      <title>Research in Progress (RIP)</title>
      <url>https://rip.trb.org/Images/PageHeader-wTitle-RIP.jpg</url>
      <link>https://rip.trb.org/</link>
    </image>
    <item>
      <title>AI Applications for Developing Contractors Prequalification Criteria and Safety Matrix for NCDOT</title>
      <link>https://rip.trb.org/View/2784615</link>
      <description><![CDATA[Different state DOTs have attempted to use safety leading indicators for heavy construction projects. The North Carolina Department of Transportation (NCDOT) has contractors’ prequalification guidelines and a formal process for constructability review meetings. However, the safety component is not much considered, which may result in disruption of construction activities due to potential site safety violations. The current prequalification process for contractors requires multiple changes to improve the contractors’ selection process and secure a safe job site. Prequalification process may be revised to incorporate a safety scoring system that considers the contractors history, Occupational Safety and Health Administration (OSHA) citations, availability of in-house safety plans and safety audits, availability of competent persons and OSHA trainers, and formal safety training attended by the contractor’s company personnel. Similarly, NCDOT formal CR meeting needs to incorporate a safety matrix to be revised by project stakeholders during the meeting. The safety matrix should tie different site activities – listed in NCDOT standard bid template – with the respective OSHA big four construction hazard. Performance-based contractor prequalification allows NCDOT personnel to evaluate contractors past performance on similar work to ensure high-quality work in future projects. Occupational safety and health criteria are somehow included in such prequalification to limit construction site accidents and violations. However, current matrix/safety prequalification criteria have significant shortfalls. In this proposed research (implementation plan), the project research team will focus on utilizing artificial intelligence (AI) and machine learning (ML) applications to decide on the attributes relevant to job site safety and the most important parameters to be used in contractors’ prequalification procedures. The research tasks will use the developed criteria through the current project phase, collect information from NCDOT, other state DOTs, and OSHA to train a machine learning model to be developed specifically to evaluate the correlation between different attributes and possible “future” performance of the contractor. Attributes include – but not limited to – contractor’s size, budget, safety practices, previous experience, employee turnover, OSHA records, and experience modification rating. Predictive analytics will provide NCDOT personnel with proper algorithm to assess/predict possible contractor’s performance. The outcomes of this research will result in seamless flow of construction site activities, reduced accidents, and avoiding arbitration and litigation. ]]></description>
      <pubDate>Wed, 30 Sep 2026 11:09:05 GMT</pubDate>
      <guid>https://rip.trb.org/View/2784615</guid>
    </item>
    <item>
      <title>Shifts in Work Arrangements and Travel Behavior: A Longitudinal Analysis of Heterogeneous Effects of Teleworking on Commuting and Non-Commuting Travel Based on Seattle Commute Surveys</title>
      <link>https://rip.trb.org/View/2780211</link>
      <description><![CDATA[Following the official end of the COVID-19 pandemic in May 2023, employer policies regarding work arrangements have continued to change dynamically. This proposed project will draw on unique data from the 2022 and 2024 Seattle Commute Surveys to analyze how shifts in work arrangements from the pandemic to the post-pandemic era heterogeneously shape employees' commuting and non-commuting travel frequencies and mode choices, using econometric models, machine learning models, and explainable artificial intelligence (AI) methods. The research team will examine individual employees' post-pandemic work arrangements and patterns of change in commuting and non-commuting travel. The research will address three primary questions: (1) How have employees' work arrangements and travel patterns shifted from the pandemic to the post-pandemic era? (2) What is the relationship between work arrangement shifts and commuting mode choices? In what ways does the relationship vary across commuting distance strata and residential location types? (3) How are work arrangement shifts connected to non-commuting travel frequency and mode choices? Are these connections heterogeneous across employees' residential locations and commuting patterns? By employing econometric and AI models, this study aims to generate critical insights for transportation demand management and policy-making.]]></description>
      <pubDate>Fri, 18 Sep 2026 10:49:13 GMT</pubDate>
      <guid>https://rip.trb.org/View/2780211</guid>
    </item>
    <item>
      <title>Characterizing Operational Disruptions in Urban Freight Networks: Patterns, Propagation, and Recovery Behaviors in Atlanta</title>
      <link>https://rip.trb.org/View/2780199</link>
      <description><![CDATA[Urban freight systems are subject to recurring disruptions that can degrade travel time reliability and operational efficiency. Despite their cumulative effects on freight movement and supply chain performance, these disruptions remain relatively understudied at the network level, particularly when only publicly funded and openly accessible data are used. This project develops a data-driven analytical framework to detect, characterize, attribute, and model recurring freight-relevant disruptions and assess their impacts on urban network performance.

The research combines truck travel times, vehicle counts and classifications, congestion imagery, crash records, and street-network data to create an integrated representation of freight system conditions. Machine learning methods are used to detect and classify disruption events, while graph-based analysis and survival modeling examine their propagation and recovery dynamics. The resulting analyses distinguish disruptions that may be amenable to public-sector intervention from those that are largely exogenous and translate these patterns into freight-relevant performance measures. A proof-of-concept analytical prototype will demonstrate how the methods can support freight system assessment and performance-based planning. The Atlanta metropolitan area serves as the testbed, and the methods are designed to transfer to other regions using comparable public data.]]></description>
      <pubDate>Fri, 18 Sep 2026 10:12:56 GMT</pubDate>
      <guid>https://rip.trb.org/View/2780199</guid>
    </item>
    <item>
      <title>Cyber Resilience of Connected and Autonomous Transportation Systems (Phase II): Game-theoretic Security Assurance of Network-level Traffic Signal Control</title>
      <link>https://rip.trb.org/View/2778265</link>
      <description><![CDATA[Two trends are transforming transportation systems. On the bright side, the increasing complexity of cyber-physical technological advances enable connectivity and autonomy through seamless integration of computation, communication, sensing, and control, holding promise for societal and economic impacts. On the dark side, however, these tightly coupled cyber-physical interdependencies can be self-defeating as they can pose new vulnerabilities to accelerating disruptions in cyber space (i.e., cyber-attacks), raising concerns about the safety, security, and privacy of the transportation system users. In view of these two overlapping trends, it is crucial to bolstering the transportation cyber-physical resilience. Building on the findings from a comprehensive literature review of transportation cyber resilience in Phase I of this project, Phase II focuses on security assurance for network-level (connected) and artificial intelligence (AI)-powered (autonomous) traffic signal control systems.

Learning-based control methods, such as deep reinforcement learning (DRL), have gained traction in traffic signal control due to their computational capability to efficiently deal with real-time changes in traffic flow dynamics. Despite growing popularity, DRL has been recently found to be vulnerable to adversarial attacks, which precludes its use in safety-critical traffic signal control applications unless its vulnerabilities are mitigated. From a methodological standpoint, mitigating the security issues facing DRL is still a daunting if not impossible task. The well-established methods in the area of (un)supervised machine learning security do not methodologically suit DRL security, since DRL deals with sequential decision-making, i.e., non-fixed distribution of training data, based on interactions with an environment. The interdependence of the current decision on the ones taken in previous time steps increases the degrees-of-freedom of adversarial attacks on DRL, e.g., an adversary can perturb multiple state observations of the environment over time.

Given the above research gap in DRL security assurance, most relevant studies resort to empirical (aka heuristic) methods, in which attack models sample only some of the possible disturbances (e.g., self-crafted perturbed data) and the devised defense strategies do not generalize out of the experienced attacks. Leaping from empirical security toward certified security, this project proposes a game-theoretic approach to robustifying DRL-based traffic signal control at the network level. The game-theoretic approach provides computational security assurance through explicitly modeling an adversary’s destabilizing perturbations rather than sampling only some of the possible perturbations. The network-level scope of the proposed project prevents higher-order cyber risks compared to isolated traffic signal control, through the explicit consideration of the underlying network effects, such as inter-agent (intersection) interactions over networks.  ]]></description>
      <pubDate>Fri, 18 Sep 2026 10:00:47 GMT</pubDate>
      <guid>https://rip.trb.org/View/2778265</guid>
    </item>
    <item>
      <title>Deep Reinforcement Learning Based Driving Strategies Bootstrapped by Suboptimal Policies</title>
      <link>https://rip.trb.org/View/2778233</link>
      <description><![CDATA[Designing autonomous driving systems requires a modular approach. This includes vehicle sensing, perception, localization, scene representation, path planning, decision-making, and vehicle control. The study focuses on decision-making in complex driving scenarios. Traditional methods often use model-based, heuristic, or rule-based controllers.  Heuristic and rule-based controllers rely on human expertise to handle different driving situations. They include safety measures and rules to avoid deadlocks. However, these methods face limitations as designing rules to cover every possible driving scenario is highly challenging and resource intensive.

Reinforcement learning (RL), on the other hand, offers a self-supervised approach, learning control policies through interaction with the environment. With advancements in neural networks, deep reinforcement learning (DRL) has shown the ability to efficiently map state-action transitions in high-dimensional environments. 

However, a persistent challenge in RL is achieving sufficient exploration of the environment. Research has shown that RL agents require extensive exploration to accurately map state-action values and discover high-reward states. Even with mechanisms like stochastic DRL algorithms that aim to embed exploration throughout the training process, agents often fail to discover optimal policies in complex scenarios. As a result, RL agents struggle with delayed rewards, where limited and inefficient exploration can result in sub-optimal performance.

Driving environments encompass a range of challenging scenarios, such as roundabouts, overtaking, merging, intersections, and highway exits. These scenarios involve delayed rewards, which discourage exploration and complicate the RL learning process. Examples of delayed rewards include the cumulative benefit of using a free lane to achieve optimal cruising speed, waiting for the right moment to enter a roundabout safely, or merging into traffic while avoiding collisions—even if it requires temporarily stopping in the lane. Failure to explore and identify these states during training can lead to sub-optimal policies or even unsafe behaviors, such as remaining stuck behind a slow vehicle, rushing into a roundabout prematurely, or misjudging merges in traffic. 

In this study, the research team proposes to address these limitations by combining the stability of rule-based sub-optimal controllers with the adaptive learning capabilities of DRL. The team aims to utilize sub-optimal policies to bootstrap efficient training during the RL learning process and to overcome exploration barriers. Specifically, the team leverages a rule-based controller to prevent deadlocks and guides RL agents toward promising states, improving their ability to learn optimal driving strategies. This approach not only improves collision avoidance but also enhances training efficiency by guiding agents through challenging scenarios where rewards are delayed. By integrating suboptimal strategies, the method strikes a balance between simplicity and adaptability, enabling agents to learn more effectively.

The experiments are conducted in a complex driving scenario where the driving policy must optimize long-term rewards. The key contributions of this study are summarized as follows:
(1) The sub-optimal policies: The team designs sub-optimal controllers capable of making driving decisions under reasonable motivations.
(2) Guidance for RL agents: The sub-optimal controllers guide RL agent toward exploring states associated with delayed rewards, improving exploration and learning efficiency.
(3) Validation in complex driving scenarios: The team validates the effectiveness of the method in complex driving scenarios, demonstrating its ability to optimize long-term rewards and handle challenging conditions effectively.]]></description>
      <pubDate>Thu, 17 Sep 2026 16:49:20 GMT</pubDate>
      <guid>https://rip.trb.org/View/2778233</guid>
    </item>
    <item>
      <title>Enhanced Crash Risk Estimation in Urban Environments: Integrating Multi-Source Data and Advanced Modeling Approaches for the City of Pittsburgh</title>
      <link>https://rip.trb.org/View/2778230</link>
      <description><![CDATA[Crash risk estimation is a critical aspect of transportation safety study, aimed at identifying high-risk locations and mitigating potential hazards/risks for road users. Limited resources can be allocated to those high-risk areas for effectively improving safety for the entire network. As travel demand grow, new mobility options emerges and traffic volumes increase, the complexity of factors influencing crash risk has also grown. Historically, crash risk estimation has been approached using statistical models, such as Safety Performance Functions (SPFs), which predict the expected crash frequency or risk on road segments based on exposure and other factors. However, the dynamic nature of urban environments and the growing availability of diverse data streams, such as video analytics, near-miss reports, real-time traffic data, and social media, present opportunities to enhance traditional methods. In particular, data directly measuring near-misses in addition to observed and rare accidents enable a more holistic and accurate measure of crash risks. These sensing and methodological advancements can improve model accuracy, reduce biases, and provide more granular insights into high-risk locations for both motorized and non-motorized traffic.

By leveraging long-term multi-source data related to crash risks, diverse sources such as police reports, traffic stops, and near-miss incidents from the city of Pittsburgh, and applying both statistical and machine learning approaches, this project aims to enhance the precision of crash risk estimations. It addresses key challenges related to data aggregation, modeling performance, and bias mitigation to provide more reliable safety insights that can inform future interventions.

Methodology and steps

The approach is to estimate SPFs for both motorized and non-motorized crash risks in the city of Pittsburgh with multiple data sources, including historical crash data, UPMC urgent care data, police traffic stop reports, 311 near miss report, 911 Dispatch data, Waze reports (social media), Velo.ai video analytics data, and so on. Specifically, the research team processes the video, medical, police and Waze data among all locations and in the past few years to either improve or validate the SPFs (namely current risk maps). The team applies both statistical and machine learning models and select the best model based on model performance. The team develops a new crash map and compare it with the the current risk map network. The team explores the potential biases and how to improve the accuracy of crash risk estimation in general. This work will be conduct particularly for the City of Pittsburgh, but can be replicated in any other communities in general. The detail for each step is listed below.
Step 1: Conduct a comprehensive literature review on risk factors and video analytics algorithm for near-miss detection
The team plans to carry out a comprehensive literature review on types of risk factors that make significant contributions to the crash risk (particularly near-misses related to vulnerable road users). This literature review helps collecting necessary datasets for building robust SPFs to estimate crash risk for streets in the following task. In addition, the team will review the approaches of detecting incidents from the video data, which could help improve the algorithm in detecting near-miss cases from the camera videos.

Step 2: Data collection and video processing for near-miss incidents
In this step, the team will collect the related datasets which could benefit the safety performance function estimation from multiple data sources. Table 1 lists potential datasets that could be applied in this research project. Furthermore, the team plans to carry out video processing with advanced deep learning techniques to detect the near-miss incidents from the camera videos from selected intersections. The result of the number of near-miss incidents and driving factors could be either used to improve the SPFs or validate the estimated SPFs.

Table 1. Source and year of the datasets
Dataset	Data source	Year
High injury network	City of Pittsburgh	-
Crash data	PennDOT	2017-2023
UPMC urgent care	UPMC	In request
Police traffic stop report	City of Pittsburgh	2018-2022
311 near miss report	City of Pittsburgh	2019-2023
911 Dispatch data	WPRDC	2015-2022
Network of bike facilities	City of Pittsburgh	-
Built environment data	EPA	-
Socio-demographic data	NHGIS	-
Waze data	Waze for Cities	2018-2024
Cyclists’ waypoints and distances to vehicles	Velo.ai	2022-2024
Camera video from selected intersections	City of Pittsburgh	In request
Strava pedestrian and bicyclist count data	Strava Metro	2023-2024

Step 3: Estimate SPFs with both statistical and machine learning modeling approaches.
The team will estimate two sets of SPFs for motorized crash risk and non-motorized crash risk, respectively, with the data collected and processed in Step 2. The team plans to apply both statistical (e.g., linear regression, Poisson regression, and Negative Binomial regression) models and machine learning models (e.g., random forest, gradient boosting decision tree, and neural network). The team will compare the model performance and select the best one as the final model.

Step 4: Develop a crash map for all street segments in Pittsburgh, compare it with the original high injury network, and prioritize high-risk locations. 
The team plans to use the estimated SPFs to predict the expected crash risk (both motorized and non-motorized) for all streets in the city of Pittsburgh and develop a crash map dashboard. The team will compare the crash risk with those estimated in the high injury network and find the differences. The team will explore the sources of bias and study how to improve crash risk estimation. Finally, the team will rank the locations and prioritize high-risk ones for further examination.]]></description>
      <pubDate>Thu, 17 Sep 2026 16:28:48 GMT</pubDate>
      <guid>https://rip.trb.org/View/2778230</guid>
    </item>
    <item>
      <title>Smart Trailer Technologies for Data-Driven Safety, Efficiency, and Policy Innovation in Heavy-Duty Fleet Operations</title>
      <link>https://rip.trb.org/View/2778229</link>
      <description><![CDATA[In 2022, 4,764 fatalities in the United States involved heavy-duty trucks, with 66% of these deaths affecting passenger vehicle occupants. Many resulted from underride collisions, where smaller vehicles slide under the high-clearance sides or rear of trailers, often causing catastrophic outcomes. Despite the severity of this issue, the transportation industry lacks a clear, standardized definition of “smart trailers” and guidelines for evaluating and mitigating such risks. This absence hinders the development of cohesive safety policies, the adoption of effective safety measures, and the creation of implementation roadmaps. Unlike Europe, where side underride guards are mandated and proven effective, the U.S. has yet to establish similar frameworks, contributing to uncertainty for policymakers, fleet managers, and technology developers.

This project addresses these challenges by clarifying the definition of smart trailers and establishing a data-driven framework to evaluate their safety, efficiency, and economic benefits. Smart trailers integrate advanced sensors, telematics, and artificial intelligence (AI) to monitor operations, predict risks, and optimize maintenance schedules. A key focus is determining the direct and indirect costs and benefits of equipping trailers with side and rear underride guards and evaluating proactive warning systems or additional technical and policy solutions to reduce trailer-related collisions. Tackling the limited real incident data, the study will integrate data-driven analysis with physics-informed machine learning to guide policymakers in developing practical, safety-enhancing regulations.

To achieve these objectives, the project will synthesize data and physics-based models from multiple sources, including accident reports, vehicle physics models, historical crash data, regulatory documents, and insurance industry records. This information will support the development of simulation-driven models to assess both the safety and economic implications of underride guards or alternative solutions. By leveraging these models, the research will provide actionable insights into the optimal design and implementation of safety measures and how existing smart trailer technologies can support evidence-based regulatory decisions.

Collaboration with key partners will ensure the research's practical relevance and impact. Clarience Technologies will evaluate current smart trailer technologies, identify limitations, and help define a standardized framework. Safety Emissions Solutions will contribute inspection and regulatory data to enhance cost-benefit analyses, while the Regional Industrial Development Corporation (RIDC) will assess how smart trailer technologies can address regional transportation challenges and resource allocation challenges. Pitt Ohio’s participation in validating the system through real-world testing and operational data collection is tentative and subject to confirmation.

The outcomes of this project include a standardized definition of smart trailers, recommendations for safety policy, and a framework for adopting advanced safety technologies. By addressing critical gaps in definitions, data, and evaluation methods, this research will guide the development of effective regulations and implementation roadmaps. Through collaboration with industry leaders, this project aims to enhance public safety, operational efficiency, and sustainable fleet management, paving the way for a safer and more technologically integrated future in heavy-duty transportation.]]></description>
      <pubDate>Thu, 17 Sep 2026 16:22:36 GMT</pubDate>
      <guid>https://rip.trb.org/View/2778229</guid>
    </item>
    <item>
      <title>UTC Safety21: Improve Transportation Infrastructure (Geotechnical Asset) Safety by Reducing the Risks of Landslides - Phase 3</title>
      <link>https://rip.trb.org/View/2778223</link>
      <description><![CDATA[Geologic hazards including slope failures, landslides, mudflows, debris flows, etc. and hydrological hazards related to floods and stormwater surge can be destructive to transportation infrastructure (geotechnical asset) and threaten property and human life along the highway, railroad and road ways. Landslides alone cause thousands of deaths and many billions of dollars in damage every year. 

Morgan State University team proposes a multi-phase (multi-year) project focusing on safety of transportation infrastructure systems by preventing geohazard, specifically slope failure and landslides and minimizing impacts of geohazard. This project will employ an integrated approach of geotechnical and Artificial Intelligence (AI)/Machine Learning methods for assessing conditions of geotechnical assets, such as cut slopes and embankment of the DOT SHA and delineating landslides and high-risk areas. 

The objectives (tasks) of the proposal include: (1) with AI/Machine Learning approaches assess the risks of landslides based on soil/rock types, weather conditions, mechanical properties of slope materials, stream gage station flow data, pavement material and design, and the status of existing retaining structures along the selected highway sections, using Maryland as case studies and railroad, (2) identify and map the high-risk areas based on controlling factors such as geometry and mechanical properties of soil or rock, and triggering factors, including gravitational and hydraulic forces, using available survey data, remote sensing and LiDAR and InSAR data and other factors like transportation modes, (3) design and test protocols for real time monitoring at selected sites in consultation with agencies and organizations (such as DOT SHA, Federal Railroad Administration), and (4) recommend strategies for reducing the risks of landslides with real-time monitoring for the high-risk areas, and improving the safety of the transportation infrastructure. All the methods and strategies can be transferred to other states or regions with similar geological conditions and engineering configurations. Phase 1 of this project covered task 1 and part of task 2. Phase 2 covers part of task 1 and task 2. Phase 3 will cover task 3 and will also include literature review and summary of landslides along the railroad. This project will primarily complement the ongoing project sponsored by the Maryland DOT (MDOT) SHA (see more information in TRID) led by Zhuping Sheng in collaboration with Carnegie Mellon University (CMU) (Dr. Sean Qian). In this phase the research team will also expand their collaboration with CMU team (Dr. Christoph Mertz) by including technology transfer in photographical images processing to build conceptual models and identify slope failures.   

As PI Dr. Sheng will coordinate the efforts in collaboration with MDOT/SHA and advise other faculty and postdoctoral research associates and graduate students to carry out the project under Tasks 1 and 2. Dr. Zhuping Sheng has experience in geohazards assessment and mitigation, geotechnical and water resources engineering.  The team includes Co-PIs, Dr. Oludare Owolabi with experience in transportation engineering and resilient infrastructure and will mainly cover Task 1 and Dr. Yi Liu with experience in geohazards, land subsidence and landslides and geotechnical engineering and will cover Task 2. They have conducted and are currently conducting research supported by MDOT SHA, which provides a strong foundation for future collaboration with the partner MDOT SHA and others for technical transfer. Dr. Kofi Nyarko, Professor of the Electrical and Computer Engineering and Director of Morgan AI/ML program, has extensive experience in AI/ML research and applications, including automating complex systems through computer vision and machine learning as well as scientific/engineering simulation & visualization, predictive visual analytics and will mainly cover Task 3.

The Morgan AI/ML program will provide in-kind support (Dr. Nyarko’s time) and support for one student who will help with sensor tests, calibration and assembling as part of technical transfer at selected sites in consultation with Maryland DOT/SHA. The team will also explore opportunity for summer intern program with both UTC and the AI/ML program's support, which is designed for development of future workforce in transportation safety. Students have participated in and will continue to participate in exchange programs and deployment partner symposium and other activities. Through this project the MSU team will continue to expand collaboration with CMU and other partner institutes via faculty meetings, seminars, national summit, and other venues, which provides great opportunity for professional development.]]></description>
      <pubDate>Thu, 17 Sep 2026 14:52:56 GMT</pubDate>
      <guid>https://rip.trb.org/View/2778223</guid>
    </item>
    <item>
      <title>Avoiding Collisions in Connected and Autonomous Driving Using Safe Deep Reinforcement Learning Exploration through Control Barrier/Lyapunov Functions</title>
      <link>https://rip.trb.org/View/2778220</link>
      <description><![CDATA[According to recent statistics in the U.S., close to 6 million motor vehicle accidents were reported in 2022 with more than 42,000 fatalities and more than 1.6 million injuries. In 2021, vulnerable road users accounted for about one-third of all traffic fatalities including 7,388 pedestrians, 966 cyclists, and 5,932 motorcyclists. In 2021, about 116,000 pedestrians, 105,000 cyclists, and 201,000 motorcyclists were treated in hospitals after collisions. Motor vehicle accidents are the second leading cause of deaths resulting from unintentional injury in the U.S. Connected and autonomous driving with carefully developed crash mitigation methods will help in the reduction of road traffic accidents and the fatalities and injuries they cause, including the ones with vulnerable road users. Based on these facts, this proposal focuses on developing data driven collision avoidance algorithms that learn to avoid collisions with nearby vehicles and vulnerable road users, building upon the research team's approach to pedestrian and bicyclist safety in their Year 1 and Year 2 Safety 21 projects. A deep reinforcement learning safety algorithm with safe exploration will be developed using realistic hardware-in-the-loop (HIL) and vehicle-in-virtual-environment (VVE) platforms.

The current state-of-the-art crash mitigation methods use a catalog of possible collision interaction scenarios using model-in-the-loop simulations for development and evaluation. While this approach will be successful in mitigating collisions for the scenarios in the catalog being used, it is not generic in nature and will not necessarily mitigate crashes in the many other possible dangerous interactions between the road actors. The motivation for the use of the data driven approach in this proposal is this limitation in scope and the difficulty of accurately modeling the interaction dynamics between vehicles and between vehicles and vulnerable road users in dangerous encounters. The deep reinforcement learning based collision avoidance algorithm will learn from interactions in complex traffic environments, adapt to various traffic scenarios, and satisfy real-time performance requirements, making it a more robust and advanced solution. The team will use a hybrid approach in that the trained deep reinforcement learning algorithm will only be activated and work during dangerous and close interactions with other road users. Training will also be based on such interactions using traffic microsimulation to generate realistic training interactions automatically as compared to the fixed nature of the scenario driven approach.

Since reinforcement learning algorithms suffer from the absence of safety guarantees during the learning process, the team will integrate their deep reinforcement learning algorithm with control Lyapunov and control barrier functions to limit its training to feasible areas of interaction with other road actors within the road barriers, with realistic vehicle dynamic model constraints as compared to the oversimplified ones used in the literature. Vehicle-to-Vehicle and Vehicle-to-VRU connectivity will also be used as low latency, and high speed near field and over the cloud cellular and other means of communication are available and as they help significantly in non-line-of-sight situations in aiding the perception sensors of autonomous vehicles. Efficacy of the proposed method will be demonstrated using HIL simulations and with a real vehicle using VVE.  ]]></description>
      <pubDate>Thu, 17 Sep 2026 13:56:00 GMT</pubDate>
      <guid>https://rip.trb.org/View/2778220</guid>
    </item>
    <item>
      <title>Completion of Hydrologic Design and Platform Data (TR-792 Phase II)</title>
      <link>https://rip.trb.org/View/2775961</link>
      <description><![CDATA[Flooding is Iowa's most persistent and costly natural hazard, responsible for approximately 80 percent of all federally declared disasters in the state since 1989. The Iowa Highway Research Board (IHRB) Project TR-792 demonstrated that On-Road Structures (ORS) are a practical tool for reducing peak flows and protecting downstream communities and infrastructure. TR-792 identified approximately 250,000 potential ORS sites statewide, developed an automated hydrologic design methodology, and deployed results through the Iowa DOT On-Road Structures (IDOT-ORS) web platform. Watershed-scale modeling in six representative HUC12 watersheds showed that strategically deployed ORS systems can reduce 50-year storm peak flows by approximately 18 percent at watershed outlets.

This Phase II proposal will build upon and expand the work completed by TR-792 by working on three interlinked objectives. Objective 1 extends the automated planning design methodology to all feasible ORS sites across Iowa, including scenario analyses that evaluate the effect of alternative roadway elevation assumptions (e.g., raising the road by one or three feet). Objective 2 develops a surrogate machine-learning model, trained on process-based hydrologic simulations, to enable rapid, large-scale estimation of watershed-scale flood reduction benefits. Objective 3 modernizes the IDOT-ORS platform, integrates all new datasets and scenario modeling capabilities, and deploys a targeted communication strategy, including educational videos, training materials, and digital brochures to translate technical outputs into actionable guidance for county engineers, watershed managers, and emergency management professionals. Similar to the approach followed for TR-792, the proposed research will be performed by a team of researchers from Iowa State University and The University of Iowa.]]></description>
      <pubDate>Thu, 10 Sep 2026 11:49:49 GMT</pubDate>
      <guid>https://rip.trb.org/View/2775961</guid>
    </item>
    <item>
      <title>Human-AI Collaborative Bridge Inspection for Advanced Damage Detection
and Defect Quantification</title>
      <link>https://rip.trb.org/View/2689757</link>
      <description><![CDATA[The aging concrete bridge infrastructure across the U.S. poses increasing risks to safety and reliability, amplifying the urgency to modernize inspection processes. Traditional manual inspection methods are labor-intensive, subjective, and inconsistent, creating barriers to effective infrastructure management. This research proposes an innovative human-AI collaborative framework designed to enhance the durability, resiliency, and technological capabilities of bridge inspections. The project integrates advanced artificial intelligence (AI) and computer vision technologies to automate damage detection and defect quantification. To ensure reliability and trustworthiness, the AI system incorporates rigorous uncertainty quantification methods, increasing transparency and inspector confidence. An active learning component strategically facilitates human-AI collaboration, enabling inspectors to focus on the most critical inspection cases identified by the AI, significantly reducing manual workloads. Additionally, engineering domain expertise, particularly in structural reliability and risk assessment, will inform and prioritize inspection processes within the AI framework. The outcomes of this project will substantially improve the accuracy, efficiency, and reliability of bridge inspections, promoting safer, longer-lasting, and more resilient transportation infrastructure. Anticipated benefits include reduced inspection costs, enhanced public safety, and a proactive approach to infrastructure management that significantly extends service life and supports sustainability. ]]></description>
      <pubDate>Thu, 10 Sep 2026 09:04:28 GMT</pubDate>
      <guid>https://rip.trb.org/View/2689757</guid>
    </item>
    <item>
      <title>Near Real-time Construction Quality Monitoring and Inspection Protocols
Using Uncrewed Aerial Vehicles</title>
      <link>https://rip.trb.org/View/2691499</link>
      <description><![CDATA[Innovative quality control technologies are used for asphalt materials to ensure smooth and long-lasting pavements. Paver Mounted Thermal Profilers (PMTPs) and Intelligent Compactors (ICs) are among the technologies used for freshly paved mat quality monitoring. However, these technologies are expensive and have some limitations in retrofitting the existing pavers and rollers from different equipment manufacturers. In the first phase of our research, aerial thermal images captured from an Uncrewed Aerial Vehicle (UAV) were used to develop a framework for hot-mix asphalt (HMA) construction quality monitoring. Thermal segregation risk was quantified using three different metrics. An object detection machine learning algorithm was implemented to develop compaction efficiency metrics that includes roller pass counts and roller speed. Various construction sites were visited in 2023 and 2024 to develop models and framework. The main goal of the second phase of the study is to extend capabilities of the models and algorithms of the aerial pavement quality monitoring protocol and improve the readiness level for implementation in real-life scenarios. The focus of the proposed work in the second phase is to develop the probabilistic density models and extend the limits of the surveyed mat. The ultimate vision is to commercialize the proposed technology as a contractor tool that can provide near real-time quality data from the ongoing construction with actionable feedback. When such data is made available to the paving crew, likelihood of achieving uniform and target in-place densities will increase and overall pavement quality will be improved. ]]></description>
      <pubDate>Thu, 10 Sep 2026 08:58:47 GMT</pubDate>
      <guid>https://rip.trb.org/View/2691499</guid>
    </item>
    <item>
      <title>AI-driven Tax Increment Financing: Revolutionizing Transportation
Infrastructure Funding</title>
      <link>https://rip.trb.org/View/2696939</link>
      <description><![CDATA[Local governments across the U.S. are facing significant challenges in funding critical transportation infrastructure improvements. While federal funding opportunities exist, communities often struggle with providing required matching funds for transportation projects. This requirement heightened the need for creative financing strategies, particularly value capture tools like Tax Increment Financing (TIF). TIF is a geographically targeted economic development tool that captures a portion of the increased property tax revenue generated by development spurred by public investments in infrastructure or other public assets and reallocates it to help finance those investments. This project introduces an innovative machine learning approach that combines advanced time series analysis with multi-factor economic modeling to generate 30-year property tax revenue forecasts. It will employ a hybrid model that integrates multiple data streams including historical land value trends, macroeconomic indicators, and planned development data. Building on TTI’s decade of expertise in TIF and value capture methodologies, this data-driven advanced AI model will enable local governments to make more informed decisions about infrastructure investments and expand the application of value capture strategies. The research aligns both Policy and Technology pillars of NCIT, leveraging artificial intelligence to enhance financial resource allocation decisions. By providing more reliable revenue predictions, this research will help reduce financial risks in infrastructure investments while promoting the broader adoption of value capture strategies. The project’s goal is to catalyze transportation infrastructure development, fostering improved mobility and economic growth across communities nationwide.]]></description>
      <pubDate>Thu, 10 Sep 2026 08:17:28 GMT</pubDate>
      <guid>https://rip.trb.org/View/2696939</guid>
    </item>
    <item>
      <title>Development of AI/ML Algorithms to Extract Information from Crash Report Diagrams for Statewide Left Turn-Related Crashes at Intersections </title>
      <link>https://rip.trb.org/View/2775490</link>
      <description><![CDATA[Left turn-related crashes at intersections remain one of the most prevalent and dangerous types of motor vehicle collisions in North Carolina and across the United States. These crashes are inherently complex, shaped by conflicting vehicle paths and diverse intersection geometries and can overwhelm a driver’s attention and decision-making processes. In 2023, nearly one-third of all fatal and serious crashes in North Carolina were attributed to left turn maneuvers, underscoring a critical and persistent safety concern. Yet, despite their significance, left turn-related crashes are routinely misclassified in police-reported data and poorly captured by existing structured datasets, obscuring the true scale and characteristics of the problem.
The proposed research addresses this longstanding data gap by developing and validating a novel AI/ML framework to extract high-value safety information from crash diagrams submitted as part of electronic crash reports. By leveraging convolutional neural networks and multi-task learning, this effort will automate the classification and characterization of left turn-related crashes at intersections—identifying key features such as intersection type, crash location, turning movements, lane configurations, and physical site elements (e.g., medians, islands). These details, often embedded in crash diagrams but absent from structured data, are essential for diagnosing intersection safety problems and formulating effective countermeasures. The research will proceed through a comprehensive, multi-phase methodology. A foundational element of this effort is the manual review and annotation of over 1,000 crash diagrams to develop a robust training and validation dataset. Using this dataset, the project team will train and test AI/ML models to accurately detect whether a crash occurred within an intersection, whether it was left turn-related, and extract additional contextual features from the diagram. 
This research fills a critical void in the current traffic safety analysis paradigm by automating a previously manual and time-intensive process. It also represents one of the first large-scale applications of AI to the interpretation of unstructured visual crash data—transforming static crash diagrams into actionable intelligence. This project supports both the North Carolina Department of Transportation’s (NCDOT) Strategic Highway Safety Plan (2024) and Highway Safety Improvement Program (2024), which prioritize intersection safety and seek to halve traffic-related fatalities and serious injuries by 2035. Upon completion, the research will yield a curated dataset of left turn-related intersection crashes, and a framework for scalable deployment across other crash types or jurisdictions. This initiative holds significant promise not only in improving the fidelity of crash data but also in enhancing the efficiency, accuracy, and responsiveness of safety analyses at the state level. The methodology developed here will serve as a foundational model for future research and applications aimed at other high-priority crash types—ushering in a new era of data-driven, AI-enabled traffic safety management.
]]></description>
      <pubDate>Wed, 09 Sep 2026 10:59:15 GMT</pubDate>
      <guid>https://rip.trb.org/View/2775490</guid>
    </item>
    <item>
      <title>Beyond Pedestrian Flow: Evaluating Urban Sidewalk Friction and Accessibility via Efficient State Space Models</title>
      <link>https://rip.trb.org/View/2775059</link>
      <description><![CDATA[Traditional sidewalk performance evaluation has primarily relied on flow metrics that treat pedestrians as uniform units. These methods typically rely on basic pedestrian counts which capture pedestrians at a single point location. Walkability metrics typically assess the suitability of a street environment for pedestrian activity based on the presence or absence of specific neighborhood or street characteristics. While useful to capture high-level neighborhood pedestrian activity, these approaches fail to capture the complex, micro-level dynamics that occur in pedestrian spaces.

To address this gap, this project introduces a comprehensive framework for evaluating sidewalk performance by developing novel friction metrics. By moving beyond simple counting, the research team aims to quantify how different pedestrian groups and infrastructural elements interact in real-world scenarios. The architecture leverages the combined strengths of Visual Language Models (VLMs) and highly efficient State Space Models (SSMs). The project is structured around three primary objectives: (1) Semantic Categorization via Visual Language Models — processing CCTV camera feeds to categorize pedestrians into distinct behavioral groups and identify infrastructural hurdles; (2) Development of SSM-Based Friction Metrics — using the semantic data generated by the VLMs, utilizing SSMs operating on anomaly detection principles to analyze specific pedestrian interactions over time; and (3) Walkability and Accessibility Evaluation — synthesizing the outputs of the pipeline to evaluate large-scale urban walkability and accessibility by creating and utilizing benchmark data.]]></description>
      <pubDate>Fri, 04 Sep 2026 15:18:47 GMT</pubDate>
      <guid>https://rip.trb.org/View/2775059</guid>
    </item>
  </channel>
</rss>