<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=PHNlYXJjaD48cGFyYW1zPjxwYXJhbSBuYW1lPSJkYXRlaW4iIHZhbHVlPSJhbGwiIC8+PHBhcmFtIG5hbWU9InN1YmplY3Rsb2dpYyIgdmFsdWU9Im9yIiAvPjxwYXJhbSBuYW1lPSJ0ZXJtc2xvZ2ljIiB2YWx1ZT0ib3IiIC8+PHBhcmFtIG5hbWU9ImxvY2F0aW9uIiB2YWx1ZT0iMTYiIC8+PC9wYXJhbXM+PGZpbHRlcnM+PGZpbHRlciBmaWVsZD0iaW5kZXh0ZXJtcyIgdmFsdWU9IiZxdW90O0RlZXAgbGVhcm5pbmcmcXVvdDsiIG9yaWdpbmFsX3ZhbHVlPSImcXVvdDtEZWVwIGxlYXJuaW5nJnF1b3Q7IiAvPjwvZmlsdGVycz48cmFuZ2VzIC8+PHNvcnRzPjxzb3J0IGZpZWxkPSJwdWJsaXNoZWQiIG9yZGVyPSJkZXNjIiAvPjwvc29ydHM+PHBlcnNpc3RzPjxwZXJzaXN0IG5hbWU9InJhbmdldHlwZSIgdmFsdWU9InB1Ymxpc2hlZGRhdGUiIC8+PC9wZXJzaXN0cz48L3NlYXJjaD4=" 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>Deep Learning–Based Digital Image Correlation for Fatigue Crack  Characterization in Steel Structures
</title>
      <link>https://rip.trb.org/View/2703927</link>
      <description><![CDATA[This proposal presents a strategic approach to improving transportation safety through the advancement of deep learning–based Digital Image Correlation (DIC) for fatigue crack characterization in steel structural components. With aging transportation infrastructure and increasing cumulative traffic loading, fatigue-related deterioration in steel bridges and related systems presents ongoing safety risks. Accurate measurement of crack-induced displacement fields is critical for reliable structural assessment and informed maintenance decisions. The primary objectives of this proposal are to advance artificial intelligence (AI)-driven DIC methods beyond the limitations of conventional correlation-based approaches by enabling sub-pixel displacement learning through synthetic data generation, incorporating physics-informed modeling of crack-induced displacement discontinuities, and supporting high-resolution analysis of large image regions without loss of spatial detail. The methodology involves grayscale synthetic speckle data generation for sub-pixel displacement learning, mechanics-based displacement field modeling using finite element simulations, and development of an attention-enhanced deep learning architecture for full-field displacement prediction. Experimental validation against commercial DIC systems will establish a transferable methodology supporting safer fatigue crack evaluation practices.
]]></description>
      <pubDate>Tue, 19 May 2026 13:48:19 GMT</pubDate>
      <guid>https://rip.trb.org/View/2703927</guid>
    </item>
    <item>
      <title>Game-Theoretical Approach for Cyberattack Modeling and Deep Learning-Based Resilience of Connected Automated Vehicles</title>
      <link>https://rip.trb.org/View/2696944</link>
      <description><![CDATA[The current state of practice in security research on connected automated vehicles (CAVs) does not consider how adversaries may evolve over time and adapt to defense strategies. This proposed research takes the state of practice well beyond the current focus on anomaly detection towards strategic response in defense strategies by developing attack defender models with game-theoretical models. Game theory provides a framework to study the strategic interactions between defenders and adversaries with conflicting objectives. Given the above background, this study will design strategic games to study attacker and defender strategies for cyber deception, as well as algorithms to compute equilibrium or optimal defense strategies in a CAV environment. Real-world data from CAV experiments conducted by PIs will be used to design game theory models in a CAV environment. The study will design two strategic games, namely a zero-sum game and a Stackelberg security game, to formalize the interactions between attackers and defenders by devising a strategic comparison between a zero-sum game and a Stackelberg security game. The proposed game models define payoff functions that capture the trade-offs between model accuracy and the success rates of attacker and defender. The dynamic attacker-defender strategies mimic real-world applications and provide the ability to provide alerts to traffic management center operators for performing cyber incident response in a timely manner, which has attracted the Virginia Department of Transportation’s (VDOT’s) interest. VDOT will serve as a partner to help with real-world implementation. ]]></description>
      <pubDate>Wed, 29 Apr 2026 11:17:14 GMT</pubDate>
      <guid>https://rip.trb.org/View/2696944</guid>
    </item>
    <item>
      <title>VISIAM (Visual Street Index for Active Mobility): An AI-Based Tool for Assessing Bikeability and Walkability of Streets</title>
      <link>https://rip.trb.org/View/2680127</link>
      <description><![CDATA[This project proposes the Visual Street Index for Active Mobility (VISIAM)—an AI-driven framework that integrates computer vision, self-supervised deep learning, and human-perception data to systematically assess bikeability and walkability at the street segment level. Current tools for assessing bikeability and walkability are limited because they rely on subjective audits or basic metrics. VISIAM addresses this gap through a multi-stage framework: computer vision models extract and classify streetscape features from Google Street View imagery, human-perception evaluations from diverse stakeholder groups rate street imagery across multiple dimensions, and the AI-derived classifications and human ratings are integrated to produce a composite bikeability and walkability score for each street segment. The final output is a citywide, street-level index visualized through interactive maps and dashboards, enabling policymakers to identify gaps, prioritize investments, and explore how infrastructure improvements could improve street quality.
]]></description>
      <pubDate>Wed, 11 Mar 2026 15:15:43 GMT</pubDate>
      <guid>https://rip.trb.org/View/2680127</guid>
    </item>
    <item>
      <title>Observational Intersection Traffic Safety Analysis</title>
      <link>https://rip.trb.org/View/2655705</link>
      <description><![CDATA[While planners and engineers design intersections with safety in mind, the intended use and actual use of facilities do not always align. This misalignment can lead to increased safety risks for all intersection participants, particularly non-motorized users including pedestrians and cyclists. Although facility utilization mismatches can be detected through observation, typical monitoring occurs only during limited peak hours, failing to fully capture comprehensive usage patterns and emerging safety concerns.

This research proposes long-term intersection monitoring to uncover emerging facility utilization patterns and assess inherent intersection safety. The approach leverages existing traffic camera infrastructure combined with modern deep learning techniques for accurate detection and tracking of vehicles, bicycles, and pedestrians. As an explicit use case, the study examines unprotected left-turns to characterize both vehicle-vehicle conflicts through time gap analysis and trajectory conflicts involving other road users. The project develops a computer vision system capable of processing trajectories to quantify left-turns with insufficient gaps, instances where vehicles fail to yield appropriately, and average time gaps, collectively providing metrics to characterize intersection safety.

This interdisciplinary project combines computer vision algorithm development expertise from the University of Nevada, Las Vegas (UNLV) with programming support from Howard University. System evaluation will occur at intersections in both the Washington, DC area and the Las Vegas metropolitan area, utilizing purpose-built high-resolution monitoring equipment for short-term deployment as well as existing lower-resolution traffic cameras for long-term analysis. The project leverages intersection equipment acquired through NSF Award Number 2216489.

Expected outcomes include research contributions in computer vision and machine learning for trajectory analysis, workforce development through student training across both institutions, and technology transfer through publications on intersection safety scoring and practitioner engagement for field deployment.]]></description>
      <pubDate>Mon, 19 Jan 2026 16:30:17 GMT</pubDate>
      <guid>https://rip.trb.org/View/2655705</guid>
    </item>
    <item>
      <title>Enhancing Airport Runway Safety through Drone-Based Inspection Systems</title>
      <link>https://rip.trb.org/View/2652212</link>
      <description><![CDATA[Kansas Department of Transportation (KDOT) aims to improve the safety and efficiency of airport runway inspections using drone technology. Currently, runway inspections are carried out through manual and vehicle-based methods, which are time-intensive, costly, and may not provide the level of detail necessary for identifying all potential safety issues. Additionally, these methods can disrupt runway operations and pose risks to inspection personnel.
Integrating high-accuracy drones equipped with imaging technology and deep learning algorithms provides a solution. By leveraging AI models for automated defect detection and classification, this approach enables KDOT to quickly identify potential hazards, quantify runway conditions, and develop a standardized health index, such as the Pavement Condition Index (PCI), for long-term maintenance planning.]]></description>
      <pubDate>Tue, 13 Jan 2026 15:04:34 GMT</pubDate>
      <guid>https://rip.trb.org/View/2652212</guid>
    </item>
    <item>
      <title>Spatio Temporal Graph Learning for Real Time Pedestrian Exposure Estimation</title>
      <link>https://rip.trb.org/View/2640189</link>
      <description><![CDATA[Pedestrian crashes occur infrequently and are often underreported, which makes it difficult for agencies to rely only on crash records when assessing safety. Traditional Safety Performance Functions do not capture short term patterns or local context, and therefore cannot fully represent changes in pedestrian activity. This project will create a new framework that uses spatio temporal graph neural networks combined with statistical modeling to estimate pedestrian exposure across different locations and time periods. The research will draw from computer vision systems, Streetlight data, manual counts, roadway characteristics, land use, and travel related factors to produce high resolution exposure estimates.

The modeling framework will include two tiers. The first tier will use generalized linear mixed models to build a baseline exposure structure, while the second tier will apply deep learning methods to capture spatial spillover effects and temporal variation such as peak periods and seasonal changes. The results will help agencies identify areas with elevated pedestrian activity and evaluate how different roadway or land use conditions influence exposure. These data will support improved pedestrian safety analysis and guide the development of timely, evidence based interventions.]]></description>
      <pubDate>Thu, 11 Dec 2025 13:45:37 GMT</pubDate>
      <guid>https://rip.trb.org/View/2640189</guid>
    </item>
    <item>
      <title>Explaining Visual Attention for Autonomous Vehicle Controllers</title>
      <link>https://rip.trb.org/View/2640186</link>
      <description><![CDATA[End to end deep learning controllers can produce strong driving performance, but their internal decision processes are difficult to interpret. This lack of clarity makes it harder for engineers to diagnose failures and can reduce public confidence in automated systems. This project will create a counterfactual explanation framework that identifies which elements in camera images, such as vehicles, pedestrians, or traffic control devices, guide actions like braking or steering. The research will apply generative video inpainting to remove or alter specific visual elements and then observe how the autonomous controller responds to these modified scenarios.

The study will integrate this method with the ADAPT architecture and evaluate it using benchmark datasets and both real and simulated environments, including the QCar testbed. The goal is to provide clear, intuitive explanations for controller decisions that support transparency and improve safety analysis. The framework will help engineers understand system behavior, locate potential weaknesses, and develop autonomous vehicle (AV) technologies that behave in ways that can be evaluated and verified.]]></description>
      <pubDate>Thu, 11 Dec 2025 13:37:48 GMT</pubDate>
      <guid>https://rip.trb.org/View/2640186</guid>
    </item>
    <item>
      <title>Evaluating the Economic and Safety Trade-offs of Interchange and Access Drive Separation Distances
</title>
      <link>https://rip.trb.org/View/2627344</link>
      <description><![CDATA[The research project will evaluate whether the Iowa Department of Transportation’s (Iowa DOT) minimum separation standards between interchanges and first access points are overly restrictive and potentially detrimental to development opportunities around those interchanges. To achieve this, the project will utilize deep learning techniques to analyze high-resolution aerial photographs to identify interchanges on state-owned roadways, their first driveway access points, and the specific aspects of development status, such as the presence of commercial or residential buildings, vacant land, or agricultural use of the surrounding land. Crash data from the Iowa dataset will be examined to assess safety outcomes about these separation distances. A critical part of the analysis will involve evaluating the economic potential of these lands and estimating the impact of separation standards on land utilization and potential economic growth. 
In addition to state-owned interchanges, the study will identify non-interchange intersections with roadways with similar AADT levels, the number of lanes, if a median is present, and other relevant geometric features to access management. The closest access point will be determined for these intersections, mirroring the approach taken with the interchanges. The crash history for these locations will be retrieved to compare the safety performance of interchanges and non-interchange intersections directly.
This analysis, focusing on interchange and access point separation distances, will help isolate the effect of these separation standards on safety and development, controlling for traffic volume and other features. By examining interchange and non-interchange sites under similar conditions, the research will determine if the minimum separation distances at interchanges are justified or could be adjusted to better balance safety with economic development, potentially informing future policy decisions. The research will also determine the amount of developable land that could be available should the standards be relaxed.
]]></description>
      <pubDate>Wed, 19 Nov 2025 14:36:56 GMT</pubDate>
      <guid>https://rip.trb.org/View/2627344</guid>
    </item>
    <item>
      <title>Physics-Informed Deep Learning and Governance Framework for Traffic Applications with Sparse Sensor Networks</title>
      <link>https://rip.trb.org/View/2606408</link>
      <description><![CDATA[This research develops a physics-informed deep learning (PIDL) framework to address critical data blind spots in transportation networks caused by limited sensor coverage, which undermines infrastructure planning and investment decisions required for federal Highway Performance Monitoring System reporting. The study combines classical traffic estimation models with data-driven deep learning to provide accurate traffic state estimation in sensor-sparse regions, helping State Departments of Transportation overcome prohibitive sensing infrastructure costs. The methodology integrates novel Fourier feature embedding algorithms to capture spatiotemporal variations, location-based trainable adjustment parameters for localized flow disruptions, and targeted collocation sampling near critical network features. The research addresses shortcomings of existing approaches where traditional physics-based models struggle with network complexity while deep learning methods require extensive data unavailable in sparse sensor environments. A collaborative pilot study with Delaware Department of Transportation will test the framework in real-world conditions with limited sensor coverage. The interdisciplinary approach brings together transportation engineers, network scientists, and public policy experts to develop both technical solutions and governance frameworks that align transportation management ecosystems with enhanced data collection capabilities for improved decision-making and infrastructure investments.]]></description>
      <pubDate>Thu, 02 Oct 2025 15:16:16 GMT</pubDate>
      <guid>https://rip.trb.org/View/2606408</guid>
    </item>
    <item>
      <title>An End-to-End Deep Learning System for Pavement Distress Detection, Severity Estimation, and Condition Reporting</title>
      <link>https://rip.trb.org/View/2578878</link>
      <description><![CDATA[Pavement condition assessment is essential for roadway asset management, yet current methods are fragmented, manual, and resource-intensive. Traditional workflows require separate tools for distress detection, severity and depth estimation, and overall condition classification, leading to inefficiencies. While deep learning models have emerged for tasks like crack segmentation or pavement condition index (PCI) prediction, most remain task-specific and lack integration. This project proposes a unified, end-to-end multitask framework using multimodal data for automated pavement assessment. By fusing high-resolution RGB images with stereo-derived depth maps, the system jointly performs distress detection, severity estimation, depth prediction, and condition classification. It also includes a natural language processing (NLP) module to generate human-readable reports tailored to agency workflows. The architecture features a shared encoder with four task-specific decoders, leveraging cross-task correlations to enhance generalization and reduce the need for separate models. Training will use a regional dataset annotated for all four outputs, with performance evaluated using intersection-over-union (IoU), mean absolute error (MAE), and F1-score. Report quality will be assessed using text similarity metrics and expert feedback. The central hypothesis is that multitask learning with multimodal inputs will improve accuracy and efficiency while reducing manual labor. This builds on the PI’s prior work in multitask PCI estimation, multimodal segmentation, and explainable reporting for infrastructure.]]></description>
      <pubDate>Thu, 24 Jul 2025 09:30:40 GMT</pubDate>
      <guid>https://rip.trb.org/View/2578878</guid>
    </item>
    <item>
      <title>Quantifying Traffic at Highway-Rail Grade Crossings Using Artificial Intelligence</title>
      <link>https://rip.trb.org/View/2573970</link>
      <description><![CDATA[Rail-highway grade crossings remain a critical safety concern, with 2,246 incidents, 266 fatalities, and 744 injuries in 2024 based on data from the US Department of Transportation (USDOT) and the Federal Railroad Administration (FRA). There is a need to explore new standards and technologies that increase safety at grade crossings and eliminate incidents. Updated reports and data from crossings are necessary to achieve that goal. However, the number of crossings in the US is vast. Although the FRA maintains a crossing inventory as a publicly available resource, which has invaluable data for operation and maintenance, many crossings have outdated information. Analysis of the data in the crossing inventory reveals that the average age of Annual Average Daily Traffic (AADT) in revisions submitted in 2024 is approximately 16.5 years old. Also, many crossings are missing that information completely. As such, there is a need to update traffic information at rail crossings. Manually counting and classifying vehicles to estimate daily traffic is tedious and labor intensive. This research intends to address the need for updated traffic information at grade crossings using Artificial Intelligence (AI). The proposed work will extend on previous research by the team by implementing advanced AI models, using Deep Learning (DL), which will automatically detect, classify, and track vehicles and pedestrians at grade crossings using video streams from cameras. The developed model will capitalize on crowd-sourced and publicly available videos for model training and validation. The model will be able to use affordable cameras to be a cost-effective solution. The AADT information produced from the model will support decision-makers to prioritize safety updates at rail crossings.]]></description>
      <pubDate>Mon, 14 Jul 2025 19:08:14 GMT</pubDate>
      <guid>https://rip.trb.org/View/2573970</guid>
    </item>
    <item>
      <title>Pavement Distress Evaluation and Cracking Indices Generation using Deep Learning</title>
      <link>https://rip.trb.org/View/2563770</link>
      <description><![CDATA[The North Carolina Department of Transportation (NCDOT) manages the nation's second-largest roadway network. To ensure safety and efficiency of this network, it is crucial to implement timely and effective maintenance strategies. This research project aims to address these needs. 

In this research, to help optimize maintenance strategies, non-crack distresses will be classified, segmented, and quantified using cutting-edge deep learning techniques. Since an on-going research project has already completed similar tasks for varying types of cracks, upon completion of this proposed study, all types (crack and non-crack) of distresses across the 14 Divisions monitored by NCDOT can be classified and quantified using deep learning models. With this approach, it is estimated that a comprehensive state-wide pavement performance assessment can be completed in one week. Consequently, the outcomes of this proposed study, combined with those from the on-going research, will enable timely updates of distress indices and Pavement Condition Rating (PCR) values. This enhanced responsiveness of NCDOT’s PMS will significantly benefit North Carolina’s roadway network in terms of durability and sustainability. In addition, specific crack metric and index, namely the Pavement Surface Cracking Metric (PSCM) and the Pavement Surface Cracking Index (PSCI), will be calculated using the ASTM E3303-21 standard. The calculated results will be highly accurate, as the length of every crack is quantified at a pixel level. Moreover, this task will standardize and enhance the reliability of crack assessments, contributing to a more effective PMS managed by NCDOT.

One potential challenge that the UNC Charlotte researchers face is identifying certain types of uncommon non-crack distresses from the raw images provided by NCDOT. The lack of training data for these distresses can directly impact the performance of the corresponding deep learning models. To address this issue, the researchers plan to work closely with NCDOT engineers to pinpoint the locations of these distresses and gather sufficient distress data for model training purposes. Another potential challenge is the time-consuming nature of the image annotation process, a common obstacle in studies utilizing deep learning techniques for image processing. Building on the experience gained from the on-going study, the researchers plan to evaluate both AI-based and self-supervised learning approaches to expedite the annotation process effectively. 

It should be noted that transferred learning from deep learning models developed in the on-going NCDOT research project will be used to develop new models in this study. This approach allows resources spent on one task to be transferred, reused, and adapted for other related tasks, significantly reducing the computational resources and time, and more importantly, leading to improved performance of newly developed models.

In summary, this research project is proposed to improve maintenance efficiency, reduce repair costs, and support NCDOT’s sustainability goals. Various approaches will be utilized to ensure the success of this project. The methods and tools developed in this project can be applied to address other challenges in the future.
]]></description>
      <pubDate>Fri, 13 Jun 2025 12:48:03 GMT</pubDate>
      <guid>https://rip.trb.org/View/2563770</guid>
    </item>
    <item>
      <title>Development of Real-time Cyberattack Prediction &amp; Warning System</title>
      <link>https://rip.trb.org/View/2548667</link>
      <description><![CDATA[With the increasing prevalence of cyberattacks targeting transportation systems, there is a critical need for a proactive framework capable of predicting potential cyber threats and issuing timely warnings. This study introduces Cyber-TFWS, a trajectory-based forecasting and warning system designed to enhance connected vehicle (CV) safety under spoofing cyberattacks. By leveraging deep learning-based forecasting models, Cyber-TFWS predicts vehicle trajectories under attack, enabling early detection and effective mitigation strategies. The research involves a detailed literature review and the implementation of a generative adversarial model (CAGAN) for trajectory prediction. Key findings demonstrate that Cyber-TFWS significantly improves traffic safety, successfully preventing 100% red-light running incidents under specific perception-reaction time (PRT) conditions. The study also highlights the role of acceleration in improving prediction accuracy and identifies challenges in forecasting trajectories with abrupt velocity changes. Extensive simulations validate the system's robustness, underscoring its potential for real-world deployment in securing intelligent transportation systems against cyber threats.]]></description>
      <pubDate>Wed, 30 Apr 2025 16:06:29 GMT</pubDate>
      <guid>https://rip.trb.org/View/2548667</guid>
    </item>
    <item>
      <title>Exploring Safety and Security Accident-Management Policies for CAVs</title>
      <link>https://rip.trb.org/View/2548664</link>
      <description><![CDATA[The adoption of self-driving Connected and Automated Vehicles (CAV) in combination with advanced vehicle technology (AVT) has been actively pursued to enhance road traffic safety and decrease the occurrence of accidents. However, despite these concerted efforts, collisions have not been eliminated. Despite in-depth exploration into the management of Accident-Management (AM-) Policies, it is evident that this exploration in isolation may not be adequate to guarantee the secure management of traffic. This inadequacy originates from the absence of a robust mechanism for enforcement. To overcome these shortcomings, this project introduces a distributed, multi-user, multi-vehicle framework for the specification, evaluation, and enforcement of AM-Policies in the context of CAVs. To this end, this project will explore user needs from the viewpoints of passengers and bystanders and define potential AM-Policies based on varying case scenarios including CAV configurations, incidents, accidents, etc. Also, the research team introduces a theoretical model for AM-Policies leveraging Attribute-based Access Control (ABAC) for policy specification, analysis, and conflict resolution. Second, the team leverage blockchain technology to establish a decentralized framework for policy management that does not necessitate dependence on a single entity within the system, utilizing smart contracts to efficiently implement autonomous and binding agreements modeling the proposed AM-Policies. Third, the team proposes a multi-modal deep model for informative and accurate decision-making, such that different types of sensor signals and visual inputs, which are needed for successfully evaluating and enforcing AM-Policies, can be effectively incorporated. ]]></description>
      <pubDate>Wed, 30 Apr 2025 15:58:11 GMT</pubDate>
      <guid>https://rip.trb.org/View/2548664</guid>
    </item>
    <item>
      <title>Strategizing for Cyber Security Enhancement Autonomous Intersection Management
(AIM) based on Multi Agent Deep Reinforcement Learning (MADRL)</title>
      <link>https://rip.trb.org/View/2543102</link>
      <description><![CDATA[This research introduces an adaptive cybersecurity framework for Autonomous Intersection Management (AIM) systems, leveraging reinforcement learning (RL) to dynamically identify, mitigate, and respond to sophisticated cyber threats. Two advanced
RL algorithms—Proximal Policy Optimization (PPO) and Advantage Actor-Critic (A2C)—were rigorously evaluated within a digital twin simulation, effectively addressing critical threats such as Denial-of-Service (DoS) attacks, malware infections, and malicious
data manipulation. Findings indicate PPO's strength in rapid early-stage adaptation, whereas A2C exhibited superior long-term stability and continuous improvement in highly complex scenarios. Reward shaping significantly enhanced performance, underscoring its role in training efficient cyber defense strategies. Ultimately, this research confirms RL’s transformative potential in enhancing the cybersecurity resilience of intelligent transportation infrastructures, setting a foundation for future integration of
hybrid RL models, adaptive reward structures, and explainable AI methodologies.]]></description>
      <pubDate>Tue, 29 Apr 2025 15:22:37 GMT</pubDate>
      <guid>https://rip.trb.org/View/2543102</guid>
    </item>
  </channel>
</rss>