<rss version="2.0" xmlns:atom="https://www.w3.org/2005/Atom">
  <channel>
    <title>Transport Research International Documentation (TRID)</title>
    <link>https://trid.trb.org/</link>
    <atom:link href="https://trid.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>Transport Research International Documentation (TRID)</title>
      <url>https://trid.trb.org/Images/PageHeader-wTitle.jpg</url>
      <link>https://trid.trb.org/</link>
    </image>
    <item>
      <title>Bayesian Network Model for River Accident Risk Prediction in Indonesia from 2007 to 2025: An SDG Approach to Water Transportation Safety</title>
      <link>https://trid.trb.org/View/2680109</link>
      <description><![CDATA[The objectives of this study are to analyze accident patterns across Indonesian inland waterways from 2007 to 2025 and to build a fuzzy Bayesian Network (BN) for risk prediction and causal mapping. The Sustainable Development Goals (SDGs), established by the United Nations as a global framework for social, economic, and environmental development, provide the policy context for this work. Our study draws on system safety and probabilistic reasoning, positioning a BN as both an empirical tool and a theoretical lens. This framework allows us to capture uncertainty, nonlinear interactions, and hidden dependencies that are often overlooked by conventional regression. We analyze 1,257 recorded river accidents using SAS Studio (version 3.8), combining descriptive statistics, inferential tests, and a Naïve Bayes classifier. We split the data set into training (70%) and testing (30%) partitions. Notably, we apply Laplace smoothing to stabilize sparse categories and evaluate accuracy via confusion matrices and area under the curve (AUC). Our analysis indicates that fatal incidents constitute 28% of total cases, with a marked increase after 2015. The BN model achieves an accuracy of 81% and an AUC of 0.84, outperforming logistic regression benchmarks. A closer look reveals that incident type and province consistently drive fatality probabilities, while seasonal patterns are surprisingly weaker than expected. This finding may reflect policy inertia in addressing high-risk routes rather than climatic cycles. Practically, the BN framework contributes to early warning systems for transport regulators. Theoretically, it refines accident modeling by integrating uncertainty and context-specific causal pathways. We demonstrate, for the first time in Indonesia, that Bayesian Network analysis meaningfully predicts river accident risks. This novelty bridges theory and practice, offering both methodological innovation and actionable policy guidance.]]></description>
      <pubDate>Wed, 17 Jun 2026 12:23:21 GMT</pubDate>
      <guid>https://trid.trb.org/View/2680109</guid>
    </item>
    <item>
      <title>An Approach to Fly-by-Wire Handling Qualities Testing Linking Development Tools to Civil Certification Requirements</title>
      <link>https://trid.trb.org/View/2713583</link>
      <description><![CDATA[Traditional certification methods may not be suitable for aircraft in the powered-lift category. These novel aircraft feature alternative fuels (e.g. electric), distributed propulsion, over-actuated control surfaces, and advanced fly-by-wire technology. A larger research portfolio focuses on developing a variety of flight test techniques for this category. One area that requires consideration for certification is handling qualities. In previous efforts, work conducted by Systems Technology, Inc. (STI), outlined an approach to using Handling Qualities Task Elements (HQTEs) for civilian certification. In addition, a catalogue of HQTEs was presented along with supporting flight test techniques. HQTEs are proposed for partial compliance demonstration. This research report extends this work further, through the introduction of the ‘Flight Test Scorecard’ (FTSC) approach. This report provides a linkage between the proposed HQTEs and certification requirements to aid the applicants in their use during the certification process. In addition, this report provides an overview of the FTSC approach intended to support the Federal Aviation Administration (FAA) and applicant to display information relating to the completion of HQTEs and any supporting data that may be shared as part of the certification process. The scope also includes revisited HQTEs, including proposed boundaries and tolerances, which were refined during flight testing conducted in this work. The objectives of this work were to review aircraft accident databases, to create the FTSC approach, to test this approach and define metrics suitable for applicants to collect as ‘Supporting Data’, and to further refine the HQTEs developed in previous work. The final demonstration of this research effort was a dedicated flight test campaign using a variable stability helicopter.]]></description>
      <pubDate>Tue, 16 Jun 2026 07:28:34 GMT</pubDate>
      <guid>https://trid.trb.org/View/2713583</guid>
    </item>
    <item>
      <title>An analysis of injury severity of inland waterway passenger vessel accidents</title>
      <link>https://trid.trb.org/View/2706174</link>
      <description><![CDATA[Bangladesh is a South Asian developing country where inland waterways are one of the popular modes of transportation. Locally available passenger-carrying vessels, such as launches, ferries, trawlers, speedboats, and country boats, are used to cross rivers and travel between regions connected by waterways. For long-distance travel, although waterways are more economical compared to roadways, this mode is considered unsafe due to frequent occurrence of severe injuries caused by accidents. To mitigate the dangers associated with inland waterway passenger vessel accidents, it is essential to identify the underlying factors contributing to injury severity. However, these factors remain largely unexplored, particularly within the context of developing countries. This study investigates the significant factors related to the injury severity of passenger-carrying vessels involved in waterway accidents. An analysis was conducted on 337 passenger vessel accidents that occurred on inland waterways between 1983 and 2017. In this study, injury severity has been categorized into three levels: no injury, injury, and fatal injury. Mixed-ordered probit regression, with variations in means and variances, has been utilized to identify the significant factors that affect injury severity. This study found that accident year, month, and time; accident route and cause; final condition of the vessel after an accident; collision type; and the qualification of the vessel master have fixed effects on passenger vessel injury severity, whereas the coefficients for accident location, vessel width, number of vessels involved, and availability of a compass vary randomly. Among these significant factors, fatal injury is mostly affected by accident location, final condition, and vessel width. These findings may help policymakers and regulatory authorities take the necessary steps to reduce the injury severity of inland waterway accidents.]]></description>
      <pubDate>Fri, 12 Jun 2026 09:19:59 GMT</pubDate>
      <guid>https://trid.trb.org/View/2706174</guid>
    </item>
    <item>
      <title>Multivariate Negative Binomial Weighted Lindley Generalized Linear Model: Methodological Innovations and Applications in Traffic Safety</title>
      <link>https://trid.trb.org/View/2712069</link>
      <description><![CDATA[The concept of a multivariate distribution is essential in statistics, allowing the simultaneous analysis of related variables. In transportation safety, such models are effective for studying crash data across multiple categories, improving predictions and evaluations of safety measures. This paper extends the negative binomial weighted Lindley (NB-WLindley) distribution, known for handling highly dispersed or sparse data, into a multivariate framework. The proposed multivariate NB-WLindley generalized linear model treats each crash category as a random variable dependent on other categories within a joint framework while preserving the marginal distributional properties. It is hierarchically defined as a mixture of NB and multivariate weighted Lindley distributions and incorporates a dependence structure to explain correlations among categories. Applications to two crash datasets show that the multivariate NB-WLindley model can simultaneously capture different crash types and severities, identifying dependencies that univariate models cannot. The study also develops a random parameters version of the model to address unobserved heterogeneity, which consistently outperforms the fixed-parameter version and yields stronger predictive performance. Overall, this work demonstrates that the proposed multivariate model offers a more flexible and accurate approach to crash analysis, enhancing the ability to capture variability and interdependence across crash categories. It provides a practical tool for improving safety assessments and supporting data-driven decision-making in transportation safety research.]]></description>
      <pubDate>Wed, 10 Jun 2026 09:06:00 GMT</pubDate>
      <guid>https://trid.trb.org/View/2712069</guid>
    </item>
    <item>
      <title>Crash root-cause identification via trace-rewarded causation chain reasoning large language model</title>
      <link>https://trid.trb.org/View/2704171</link>
      <description><![CDATA[Road traffic crash is one of the top leading causes of death worldwide. To support the deployment of modern safety improvements such as Highly Automated Driving (HAD), collision analysis requires to reveal crash formation and find the root-causes. This study explores Large Language Model (LLM)-based techniques to reconstruct the crash causation chain, and reveal the root-causes underlying crash occurrence. By designing trace-reward functions, a domain reasoning model base upon DeepSeek-R1-Distill-Qwen-1.5B is constructed to identify the root-causes. Specifically, the trace-rewards are constructed from accuracy in recognizing crash types and entities, accuracy in extracting crash-related behaviors, and alignment degree of behavior and root-cause. Monte Carlo Tree Search (MCTS) is then employed to broaden the exploration of potential root-causes and build inference paths. Finally, Group Relative Policy Optimization (GRPO) is applied to identify the optimal inference traces through training the model. Empirical analyses are conducted using Multi-Modal Accident Video Understanding (MM-AU) dataset. The results show that the proposed method raises the Micro Accuracy of root-cause identification from 0.427 to 0.870 and improves the Macro Recall from 0.389 to 0.852. This demonstrates that the proposed method effectively enhances the benchmark model’s ability to understand the process of crash formation. Based on the identified root-causes, further analyses and discussions are conducted, providing effective support for traffic safety management and the development of preventive strategies.]]></description>
      <pubDate>Thu, 04 Jun 2026 11:56:52 GMT</pubDate>
      <guid>https://trid.trb.org/View/2704171</guid>
    </item>
    <item>
      <title>Dynamic evolution of maritime accidents: Analysis of secondary accidents using large language models and data-driven dynamic Bayesian networks</title>
      <link>https://trid.trb.org/View/2702275</link>
      <description><![CDATA[This study proposes a novel approach to model the dynamic evolution of maritime accidents, integrating a semantic extraction large language model (SE-LLM) and a two-slice Dynamic Bayesian Network (DBN). The proposed approach extracts semantic information from unstructured accident investigation reports to construct a causal accident chain model using large language models (LLMs), which quantifies the progression from primary accidents (PAs) to secondary accidents (SAs) and ultimately to final consequences. A two-slice DBN is constructed to model the temporal evolution of accidents, while sensitivity analysis is conducted to identify key risk factors. Real-world maritime accident data are used to demonstrate the effectiveness of the proposed approach in proposing effective safety enhancement strategies. Findings show that the SA type is highly related with the PA type. Furthermore, the key risk influencing factors (RIFs) for maritime secondary accident evolution are gross tonnage, vessel type, and visibility. These findings provide targeted rescue planning and resource deployment strategies for maritime rescue authorities to reduce losses caused by maritime accidents.]]></description>
      <pubDate>Thu, 04 Jun 2026 11:56:25 GMT</pubDate>
      <guid>https://trid.trb.org/View/2702275</guid>
    </item>
    <item>
      <title>Hybrid analysis of causal factors of marine accidents in ports</title>
      <link>https://trid.trb.org/View/2707617</link>
      <description><![CDATA[Marine accidents in port areas pose significant risks due to increasing vessel traffic and operational complexity. These incidents can cause economic losses, environmental damage, and serious consequences for human life. Therefore, analysing their causes and developing preventive strategies is essential. This study proposes a hybrid approach integrating the Human Factors Analysis and Classification System (HFACS) with Bayesian Networks (BN) to examine the causes of port accidents. Based on the analysis of 150 accident reports, the model classifies human errors and organizational deficiencies and probabilistically models their causal relationships. Findings indicate that decision-based errors, rule violations, authority abuses, and procedural deficiencies are critical factors, while unsafe operational conditions frequently trigger accidents in port operations. Unlike previous studies in aviation and maritime domains, this research extends the HFACS framework into a five-level structure reflecting the high-density, spatially constrained, and multi-stakeholder nature of port environments. The additional layer represents operational constraints in port waters as a distinct causal level. Conditional probability tables are developed using a Fuzzy-SAM-based expert elicitation approach rather than conventional frequency-based estimation or simple expert averaging. This method models expert consensus under uncertainty and strengthens BN parameterization. The proposed framework supports understanding port accident causation and developing effective preventive strategies.]]></description>
      <pubDate>Mon, 01 Jun 2026 09:14:40 GMT</pubDate>
      <guid>https://trid.trb.org/View/2707617</guid>
    </item>
    <item>
      <title>Modeling the Point of Derailment and Derailment Severity of Freight Trains Using Markovian Dependence</title>
      <link>https://trid.trb.org/View/2673475</link>
      <description><![CDATA[Train derailments are a major safety concern in rail transportation, often leading to substantial economic losses and risks to human life. Traditional derailment analyses typically treat derailments as independent events with uniform probabilities along a train, which fails to capture the sequential dependencies seen in real derailment cascades. This paper introduces a car-level Markov modeling framework that explicitly represents dependencies between adjacent cars to predict both the point of derailment (POD) and the subsequent propagation along the train consist. Using U.S. Federal Railroad Administration freight-train derailment data from 2011 to 2023, model parameters are estimated through a likelihood-based approach with LASSO regularization to identify the most influential operational and positional factors while preventing overfitting. The proposed model outperforms independence-based benchmarks and reveals distinct mechanisms for derailment initiation and propagation; in particular, the derailment state of the preceding car is the dominant determinant of conditional derailment risk. The framework yields interpretable, car-specific risk estimates that can support risk-based inspection scheduling, consist-design optimization and preventive maintenance planning.]]></description>
      <pubDate>Mon, 01 Jun 2026 09:13:31 GMT</pubDate>
      <guid>https://trid.trb.org/View/2673475</guid>
    </item>
    <item>
      <title>E-Scooter Riding Behaviors and Risks from Naturalistic Driving Study and Crash Data Analysis</title>
      <link>https://trid.trb.org/View/2579095</link>
      <description><![CDATA[E-scooters are becoming increasingly popular as a convenient, fun, and environment-friendly micro-mobility option, especially among younger generations. As the numbers of e-scooter riders increase across cities and towns, related crashes and injuries increase at the same time. This chapter presents a summary of the major results and findings obtained from a multi-year, large-scale naturalistic study that addresses the following important issues: (1) Baseline moving patterns of e-scooters in diverse road environments and location; (2) interaction of e-scooter riders with vehicles and other road users in different scenarios; and (3) the common scenarios for crashes or near-misses involving e-scooter riders. The collected data and analytical results can be used to develop behavior prediction models for e-scooter riders which will help support the development of automated driving systems in challenging urban environments to improve road safety.]]></description>
      <pubDate>Thu, 28 May 2026 17:09:37 GMT</pubDate>
      <guid>https://trid.trb.org/View/2579095</guid>
    </item>
    <item>
      <title>Micromobility Safety</title>
      <link>https://trid.trb.org/View/2580763</link>
      <description><![CDATA[Personal Micromobility Devices (PMDs), which are micro-sized and have limited power and speed, are a growing industry gaining popularity worldwide. Although PMDs are available for purchase, in recent years, the rise of shared e-scooter and e-bike service providers has made these devices widely used. Especially preferred for short-distance urban travel, e-scooters and e-bikes have some advantages such as better access to public transportation, less impact from traffic congestion, more economical travel opportunity for short distances, easy access to devices, and not creating air and noise pollution. However, there are also some disadvantages such as the danger they pose to pedestrians with any sight or mobility impairments by e-scooters parked/left on the sidewalk and ride on the sidewalk, and the disruption they create in traffic flow due to their lower speeds. In recent years, we have witnessed an increase in accidents involving e-scooters and e-bikes in many cities around the world. However, since these devices are not yet identified in official police records in many countries, it is difficult to access accident statistics. Lack of physical protection around these devices, particularly in collisions with motor vehicles, has resulted in serious consequences such as death and serious injury for e-scooter and e-bike riders. The low rates of helmet use among e-scooter riders also contributed to the severity of injuries. There are many risk factors that influence crashes involving PMDs. Excessive speed is a very important risk factor contributing to e-scooter and e-bike accidents. Many studies show that riders under the influence of alcohol and drugs are more likely to be involved in accidents and to be seriously injured. This section of the book provides statistical information on e-scooter and e-bike accidents and information on risk factors affecting accidents. It also provides recommendations for decision makers and micromobility operators.]]></description>
      <pubDate>Thu, 28 May 2026 17:09:37 GMT</pubDate>
      <guid>https://trid.trb.org/View/2580763</guid>
    </item>
    <item>
      <title>Model-Based Generation of Representative Rear-End Crash Scenarios Across the Full Severity Range Using Pre-Crash Data</title>
      <link>https://trid.trb.org/View/2617761</link>
      <description><![CDATA[To quantitatively estimate the safety impact of driving automation systems through simulation, it is crucial to use representative baseline pre-crash scenarios. However, such baselines generated through existing methods are generally biased towards either non-severe or severe crashes, as the underlying data used are biased. This study sought to address this issue by combining rear-end pre-crash kinematics data from naturalistic driving and in-depth crash data to create a representative dataset of rear-end crash characteristics across the full severity range in the United States. Multivariate distribution models were built for the combined dataset, and a driver behavior model for the following vehicle was created by combining two existing models. Simulations were conducted to generate a set of synthetic rear-end crash scenarios, which were then weighted to create a representative synthetic rear-end crash dataset. Finally, the synthetic dataset was validated by comparing the distributions of parameters and the outcomes (Delta-v, the total change in vehicle velocity over the duration of the crash event) of the generated crashes with those in the original combined dataset. The synthetic crash dataset can be used for the safety impact assessments of driving automation systems and as a benchmark when evaluating the representativeness of scenarios generated through other methods.]]></description>
      <pubDate>Thu, 28 May 2026 17:09:36 GMT</pubDate>
      <guid>https://trid.trb.org/View/2617761</guid>
    </item>
    <item>
      <title>From narratives to probabilistic quantification: Predicting and interpreting drivers’ hazardous actions in crashes using large language model</title>
      <link>https://trid.trb.org/View/2705448</link>
      <description><![CDATA[Vehicle crashes involve complex interactions between road users, split-second decisions, and challenging environmental conditions. Among these, two-vehicle crashes are the most prevalent. Accounting for approximately 70% of roadway crashes, they present a significant challenge to traffic safety. Identifying Driver Hazardous Action (DHA) is essential for understanding crash causation, yet the reliability of DHA data in large-scale databases is limited by inconsistent and labor-intensive manual coding practices. Here, we present an innovative framework that leverages a fine-tuned large language model (LLM) to automatically infer DHAs from textual crash narratives, thereby improving the validity and providing quantifiable probabilistic insights into DHA classifications. Using five years (2019–2023) of two-vehicle crash data from Michigan Traffic Crash Facts (MTCF), we fine-tuned the Llama 3.2 1B model on detailed crash narratives and benchmarked its performance against conventional machine learning classifiers, including Random Forest, XGBoost, CatBoost, and a neural network. The fine-tuned LLM achieved an overall accuracy of 80%, surpassing all baseline models and demonstrating pronounced improvements in scenarios with imbalanced data. To increase provide quantifiable probabilistic insights, we developed a novel probabilistic scenario analysis approach, analyzing model output shifts across original test sets and three targeted counterfactual scenarios: variations in driver distraction and age. Our analysis revealed that introducing distraction for one driver substantially increased the likelihood of “General Unsafe Driving”; distraction for both drivers maximized the probability of “Both Drivers Took Hazardous Actions”; and assigning a teen driver markedly elevated the probability of “Speed and Stopping Violations.” Together, our framework and analytical methods provide a robust and quantifiable solution for large-scale automated DHA detection, offering new opportunities for traffic safety analysis and intervention.]]></description>
      <pubDate>Thu, 28 May 2026 17:09:30 GMT</pubDate>
      <guid>https://trid.trb.org/View/2705448</guid>
    </item>
    <item>
      <title>Speed Determination Using Audio Analysis of Dash Camera Video from Commercial Vehicle Tires Frequencies for Vehicle Accident Reconstruction</title>
      <link>https://trid.trb.org/View/2692108</link>
      <description><![CDATA[Prior research has validated a reliable method for determining vehicle speed using audio recorded by cameras mounted in vehicles, specifically for rolling passenger vehicle tires. Passenger vehicle tires produce a frequency component directly correlated to vehicle speed when traveling on concrete roadways. However, prior research has not been conducted on audio for rolling commercial vehicle tires, which differ in construction from passenger vehicle tires. The stiffer Commercial tires produce audio signals on roadway surfaces that passenger vehicles tires did not when tested in the prior study. The current research concluded that commercial vehicle tires rolling on various roadway surfaces also generated a frequency that varied with vehicle speed. The purpose of this study was to outline, test, and confirm the source of the speed-dependent frequency and to develop a validated method for use in forensic applications. A modified version of the passenger vehicle tire equation from prior research was developed to account for the difference in construction associated with commercial vehicle tires, which takes into account only the outer or inner treadblocks whereas the passenger vehicle equation took both into account. Data collection included a commercially available dash camera, a GoPro camera, and a pair of remote microphones to identify the source of the speed-dependent frequency. A baseline speed was captured using GPS data from a Racelogic VBOX Sport. iZotope RX was used to identify and compare the speed-dependent frequencies of interest with the baseline speed. The proposed audio analysis method is similar to the prior audio analysis method(s) to determine vehicle speed. The proposed method can be useful in evaluating other available evidence for use in a vehicle for accident reconstruction analysis.]]></description>
      <pubDate>Thu, 28 May 2026 17:09:30 GMT</pubDate>
      <guid>https://trid.trb.org/View/2692108</guid>
    </item>
    <item>
      <title>Guidance for Left-Turn Flashing Yellow Arrow (FYA) Implementation in Nebraska</title>
      <link>https://trid.trb.org/View/2703931</link>
      <description><![CDATA[This research evaluates the safety performance of flashing yellow arrows (FYA) and driver behavioral responses in Nebraska, using data from 324 FYA intersections (Lincoln 160; Omaha 164) to develop evidence-based implementation guidance. The study addresses critical knowledge gaps in local driver acceptance patterns, the effectiveness of phasing configurations, and a comprehensive safety assessment that integrates crash and conflict analyses. The methodology employed four complementary analyses: (1) negative binomial crash frequency modeling of 3945 unique left turn crashes (2015-2024) across Lincoln and Omaha; (2) binary logistic regression of 948 gap acceptance decisions across 43 intersections; (3) linear regression of post-encroachment time for 613 completed left turns; and (4) detailed video investigation of 18 crashes at three Omaha intersections with lead-lag FYA phasing. Results demonstrate no statistically significant overall crash increase post-FYA installation when controlling for exposure (Lag: Incidence Rate Ratio [IRR]=0.937, p=0.364; Lead: IRR=1.027, p=0.563), though aggregate trends were influenced by five high-volume outlier intersections. Sensitivity analysis excluding outliers revealed lag phasing produced a statistically significant 15.1% reduction in crashes (IRR=0.849, p=0.038). Perceived Yellow Trap" (PYT) phenomenon, where lead-lag phasing configurations created perceptual confusion during phase transitions, accounting for 72% of observed crashes. Gap acceptance analysis showed lag phasing associated with 10% shorter critical gaps (3.85s vs. 4.28s), enabling higher operational efficiency. Recommended operational thresholds include prioritizing lag phasing at high-exposure locations, refining exposure thresholds using cross-product metrics, optimizing signal timing, and time-of-day operation. When properly implemented and following the recommended operational thresholds, FYA installation should improve intersection safety.]]></description>
      <pubDate>Thu, 28 May 2026 16:15:52 GMT</pubDate>
      <guid>https://trid.trb.org/View/2703931</guid>
    </item>
    <item>
      <title>A Novel AI-Powered Approach to Derive Minimum Dynamic Time-to-Collision Using Vehicles’ Key Conflicting Points: A Point-Level Framework for Traffic Conflict Analysis at Signalized Intersections</title>
      <link>https://trid.trb.org/View/2705423</link>
      <description><![CDATA[Traffic conflict analysis at signalized intersections provides a proactive approach for evaluating roadway safety without relying solely on historical crash data. However, existing computer vision–based methods often rely on simplified vehicle representations and static conflict indicators, which limit their ability to capture continuously evolving vehicle interactions. This study proposes a traffic conflict identification framework that utilizes a deep neural network–based pose estimation algorithm to extract vehicle key points from surveillance camera footage and transform them into a plan-view representation. Using these reconstructed vehicle polygons, a new surrogate safety measure, minimum dynamic time-to-collision (mDTTC), is developed to continuously evaluate vehicle interactions at the point level. Unlike traditional time-to-collision (TTC), the proposed metric accounts for dynamic motion states, vehicle orientation, and evolving interaction geometry over time. The framework was applied to a set of conflict scenarios at signalized intersections using multicamera video data. The results demonstrated that traditional TTC frequently underestimated or misrepresented conflict severity, particularly in head-on and angle conflicts, leading to false severe conflict indications. In contrast, the proposed mDTTC metric provided more accurate conflict characterization by continuously updating vehicle dynamics and spatial relationships.The proposed approach enhances the accuracy of traffic conflict detection and severity assessment while offering a scalable and cost-effective solution using existing surveillance infrastructure. The framework supports improved intersection safety evaluation and enables more reliable identification of safety-critical events.]]></description>
      <pubDate>Tue, 26 May 2026 16:57:27 GMT</pubDate>
      <guid>https://trid.trb.org/View/2705423</guid>
    </item>
  </channel>
</rss>