TRB Publications Index

Differentiated Service Pricing Strategies for Ride Hailing Considering Rational Inattentive Passengers

Tue, 19 May 2026 16:51:48 GMT

Considering the ambiguity of passengers’ perception of the real service level of ride hailing, this paper establishes a bi-level optimization model for the pricing strategies of differentiated services on ride-hailing platforms based on the theory of rational inattention. It explores the pricing mechanism of ride-hailing platforms for two optimization objectives (profit maximization and social welfare maximization) under three service strategies (providing both high and low differentiated services, providing a single low service, and providing a single high service). The model is solved using a combination of the particle swarm optimization (PSO) algorithm and the method of successive averages (MSA). Numerical examples have verified the effectiveness and robustness of the model and algorithm. The results show that under three strategies, the platform profit shows a trend of first increasing and then decreasing with the increase in the unit information cost. To ensure maximum profit, the platform should disclose the service information of ride hailing as much as possible but also maintain the unknownness of ride-hailing services appropriately. When the goal is to maximize social welfare, when the unit information cost is low, compared with providing a single service strategy, a differentiated service strategy can better meet the differentiated choices of drivers and passengers, thereby ensuring higher social welfare. As the unit information cost increases, the platform needs to reduce the service prices to attract rational inattentive passengers to choose ride-hailing services, thereby improving social welfare. The results can provide references for ride-hailing platforms in formulating pricing strategies for differentiated services.

Research on Bridge Surface Damage Classification Based on Improved Vision Transformer Model

Tue, 19 May 2026 09:02:16 GMT

To improve the accuracy of surface damage classification for concrete bridges, this paper proposes an improved model—convolutional neural network-vision transformer (CNN-ViT). First, by replacing the original image block operation with a CNN, the model’s feature extraction capability is enhanced, allowing it to retain more critical information from the image. Second, the introduced local aggregation module dynamically focuses attention on the damaged area. By aggregating local features and fusing contextual information, it enhances feature learning and extraction in the damaged region, thereby improving the model’s accuracy and robustness in identifying fine damage in complex backgrounds. Finally, to verify the model’s effectiveness, ablation experiments were conducted, and its performance was compared with that of other neural network models. Experiment results show that the model achieves an accuracy of 98.7% in real-world concrete bridge surface damage identification, which is 10% higher than that of the original model. Compared with other neural network models, the combination of CNN and the local aggregation module effectively suppresses background noise interference and significantly improves the model’s overall performance, with higher detection accuracy and robustness.

Crack Performance Modeling of Mixtures and Pavements with Consideration of Aging, Moisture Conditioning, Climatic Zone, and Reclaimed Asphalt Pavement Mitigation Strategies

Tue, 19 May 2026 09:02:16 GMT

Pavement performance prediction is essential for developing durable asphalt mixtures that meet long-term service requirements. With growing interest among transportation agencies in maximizing the use of recycled asphalt materials (RAM) to promote sustainability and cost-effectiveness, challenges persist in ensuring adequate durability of high-RAM mixtures. This study, conducted as part of the NCHRP Project 09-65, aimed to enhance RAM utilization while maintaining performance standards related to cracking resistance and durability. Six robust asphalt mixtures—representative of two distinct climatic zones and designed using various high-RAM mitigation strategies—were selected based on their performance in laboratory assessments. These mixtures were evaluated using two mechanistic pavement modeling approaches: a fracture mechanics-based cohesive zone model and a continuum damage mechanics-based FlexPAVE™. A representative pavement structure of the Federal Highway Administration’s Pavement Testing Facility and laboratory test results of mixtures formed the inputs for the models. Results from both modeling approaches confirmed that RAM mitigation strategies such as the use of recycling agents, polymer-modified asphalt, and reduced RAM binder availability can significantly affect early- and long-term resistance to cracking. Although performance prediction and rankings differed slightly between the two pavement modeling methods, the combined approach offers a mechanistically grounded framework for evaluating and optimizing high-RAM asphalt mixtures for durable pavement structure.

Investigating Nonmotorist Crash Exposure at Highway–Rail Grade Crossings Using Artificial Intelligence-Based Object Detection and Generalized Linear Count Models

Tue, 19 May 2026 09:02:16 GMT

A critical aspect of crash prediction models for highway–rail grade crossings (HRGCs) is crash exposure, which is a measure of train and highway traffic. Although data on motor vehicle traffic (e.g., annual average daily traffic) and train traffic at HRGCs are invariably available, nonmotorist traffic data at HRGCs are not readily available. Current Federal Railroad Administration and other HRGC crash models focus on train and motor vehicle traffic, overlooking nonmotorized traffic. Therefore, there is a need to gather nonmotorized traffic data to improve HRGC crash prediction models. To address this gap, nonmotorist traffic video data were recorded in this study at various urban and suburban HRGCs in Nebraska, followed by the application of an artificial intelligence-based You Only Look Once (version 8) algorithm for automated nonmotorist traffic volume detection. Data on HRGC characteristics, including surrounding area population density and land use, were collected to create a comprehensive HRGC safety database for nonmotorists. Three negative binomial models were estimated to analyze pedestrian, bicyclist, and combined nonmotorist exposure in relation to daily volumes, utilizing physical, dynamic, and temporal characteristics of HRGCs. Results indicated that sidewalks, greater visibility, and cloudy weather conditions were associated with increased nonmotorist traffic volume. Conversely, higher vehicular traffic levels, wet road conditions, low population density, and more traffic lanes correlated with lower nonmotorist traffic. This study established an initial framework for nonmotorist traffic monitoring and identified key environmental and technical challenges in automated detection at HRGCs; based on these findings, recommendations for addressing technical limitations were provided for future research.

Event Prediction Large Language Model: Prediction Framework for Highway Abnormal Events Based on Large Language Models

Tue, 19 May 2026 09:02:16 GMT

In response to the common issues of fragmented spatiotemporal feature modeling and insufficient interpretability of deep learning models in highway network anomaly event prediction, this paper proposed the event prediction large language model (EP-LLM) framework based on LLMs. The framework employs structured spatiotemporal prompt engineering and parameter-efficient fine-tuning techniques to optimize both prediction accuracy and decision transparency. To address these challenges, we constructed a spatiotemporally aligned multichannel anomaly event dataset to overcome single-source data limitations. We then applied low-rank adaptation to directionally fine-tune the large language model’s decoding layer, enabling multidimensional predictions of congestion levels, accident types, and high-risk intrusions. In addition, a structured prompt template was designed to encode the periodicity, trend, and spatial hotspots of event frequency into interpretable semantic vectors, which was combined with a chain-of-thought reasoning architecture to enhance accuracy. Experimental results demonstrated that EP-LLM substantially outperformed traditional models (long-short term memory [LSTM], light gradient-boosting machine, temporal convolutional network) and closely approaches the accuracy of closed-source generative pretrained transformer (GPT)-4o. Specifically, it achieved a 25.4% reduction in mean absolute error (MAE) against LSTM and a 47.3% decrease compared with full-parameter fine-tuning—the latter confirming the substantial advantage of our parameter-efficient design. Ablation studies verified that the synergistic integration of low-rank fine-tuning and structured spatiotemporal prompting collectively accounts for over 47% of the MAE improvement. For multistep forecasting (3–9 steps), the framework reduced error fluctuation by 32.3% while maintaining a stable mean absolute error-root mean square error dispersion coefficient (0.32), demonstrating enhanced robustness. Crucially, EP-LLM delivers these advances with 43% lower computational resource consumption than GPT-4o, achieving an optimal accuracy-efficiency trade-off.

Evaluating the Influence of Stop Line Position on Driver Stopping Distances at Rural, Stop-Controlled Intersections

Mon, 18 May 2026 14:04:39 GMT

Rural, two-way, stop-controlled intersections represent a high risk for serious injury and fatality owing to the increased propensity for right-angle crashes following stop sign running or poor gap acceptance by drivers on minor roads. Safety countermeasures are often aimed at improving minor road stop sign compliance, gap selection, sight distance, and major road speeds. Stop lines have been found to increase intersection recognition and indicate stopping position. A simulation study of licensed drivers (n = 50) examined stopping behaviors at a random series of two-lane, two-way, stop-controlled intersections with no stop line, a stop line 1.22-m back from the lane edge, or 9.14-m back from the intersection edge. Participants experienced each stop line type twice. Stopping distances were observed at two slow speed/stopping thresholds: <2.0 m/s and 0.0 m/s. Of the three conditions, participants stopped closest to the intersection with the 1.22 m stop line and furthest from the intersection with no stop line. At both stopping speed thresholds, differences in stopping distances were statistically significant between the 1.22 m stop line and the 9.14 m stop line, the 1.22 m stop line and no stop line, and between the 9.14 m stop line and no stop line. Order and first/last stopping distance differences were nonsignificant. The results suggest that stop lines may perceptually anchor drivers to stop closer to intersections and may help increase sight distances, improve gap acceptance, and slow mainline driver speeds.

How Do Changes in Efficiency Affect Road Transportation Carbon Emissions? Efficiency-Biased Production-Theoretical Decomposition Analysis

Mon, 18 May 2026 14:04:39 GMT

Amid global climate mitigation efforts, China’s road transportation sector faces escalating tensions between persistent transportation scale growth and stringent decarbonization commitments. Existing studies have shown that efficiency improvement—defined as the ability to maximize transport output while minimizing resource inputs and reducing undesirable outputs—is one of the important means to achieve carbon emission reduction. However, the quantitative contributions of efficiency improvement in the change of road transport carbon emissions (RTC) still lack a systematic evaluation. This study pioneers an efficiency-biased production-theoretical decomposition analysis framework to investigate the role of efficiency in RTC changes across 30 Chinese provinces during 2010–2020. Key empirical findings include: (1) total factor production efficiency (TFPE) in road transportation increased from 2010 to 2018 but receded amid the pandemic, exhibiting regional disparities—the eastern region registered the lowest TFPE (e.g., Beijing and Shanghai < 0.6), whereas the central region (e.g., Hebei = 1) achieved optimal performance; (2) the expansion of the transportation scale constituted the principal impetus for RTC escalation, while potential energy intensity and TFPE emerged as key mitigating forces—among factor-specific efficiency effects, except for transport output biased efficiency effect, other factor-biased efficiency effects all contributed to reducing RTC; and (3) the potential for efficiency-driven emission reduction varied by region—capital-biased efficiency predominated in the west (32.9% annual average), labor-biased efficiency in the east (33.4%), while energy and carbon-biased efficiencies exhibited limited potential because of technological entrenchment. These results emphasize the need for region-specific policies: western China should prioritize green infrastructure upgrades, while eastern regions should leverage intelligent transport systems and labor training. The findings provide empirical support for optimizing emission control frameworks and inform strategies to harness efficiency potential for achieving China’s dual carbon goals.

Revised Faulting Model to Account for Dowel Looseness and Corrosion in Long-Life Pavements

Mon, 18 May 2026 14:04:39 GMT

The current AASHTOWare Pavement Mechanistic-Empirical Design faulting model framework has several limitations, which prohibit the consideration of alternative dowels commonly used in long-life pavements. Users cannot account for key design parameters, such as dowel stiffness, which is a critical need given the increased use of alternative dowel bars. Also, the effect of corrosion is not integrated into the model. Lastly, because of a lack of available faulting data from doweled pavements, the calibration alone is unable to account for the effect of loss of dowel performance resulting from corrosion. This study presents a revised faulting model framework which incorporates key design, loading, and environmental parameters. First, a comprehensive dowel damage model was developed based on an accelerated dowel loading test. The damage model incorporates critical parameters such as dowel stiffness which, before this work, could not be directly considered. Second, a novel corrosion model informed by a laboratory analysis is incorporated to account for the reduction of dowel diameter caused by corrosion. Lastly, the concept of “equivalent dowel diameter” was introduced into the faulting model. The faulting model was calibrated using faulting data from a national database of in-service pavements. A series of model adequacy checks was conducted to demonstrate that the model does not exhibit bias and to illustrate the effect of key parameters on predicted faulting. The improved faulting model framework is the first comprehensive model able to account for damage accumulation resulting from both vehicle loads and corrosion for the range of dowel bars currently on the market.

Driving Behavior and Instantaneous Emissions in Extra-Long Expressway Tunnels: PEMS Measurements and Machine-Learning Analysis

Mon, 18 May 2026 14:04:39 GMT

Extra-long tunnels, characterized by sudden changes in lighting and spatial configuration, exert a considerable effect on driving behavior, fuel consumption, and vehicular emissions. Despite this, investigations into the interrelationships among these factors in such environments remain scarce. In this study, driving behavior and instantaneous emissions of CO, CO₂, and NOx in extra-long expressway tunnels were investigated using a portable emission measurement system integrated with vehicle operational sensors. A light-duty gasoline vehicle was tested across entrance, mid-tunnel, and exit sections of the Qinling and Li Jia He 3# tunnels in China. The experiment was conducted over five consecutive days, with a different designated driver completing one round trip (covering both tunnels) each day. Results showed that the entrance section exhibited frequent acceleration, highest power demand, and peak CO or CO₂ emissions. Emission peaks were closely associated with periods of unstable acceleration and deceleration, particularly at tunnel transition zones. High-emission events for CO and CO₂ predominantly occurred at 60–70 km/h, with slight acceleration (0–0.5 m/s²). Furthermore, the study underscores the potential of machine-learning models for predictive emission analysis. Machine-learning models (CatBoost for CO, Random Forest for CO₂ or NOx) predicted emissions with test-set R² values of 0.637 (CO), 0.557 (CO₂), and 0.206 (NOx), indicating moderate predictive capability for CO and CO₂ but limited performance for NOx under the tested conditions. These results offer empirical support for optimizing tunnel design and traffic management strategies aimed at reducing emissions and enhancing safety, contributing valuable insights toward the development of sustainable transportation infrastructure.

Traffic Flow Impacts of Autonomous Utility Service Vehicles Representing Low-Speed Operation in Urban Networks

Mon, 18 May 2026 14:04:39 GMT

Autonomous utility service vehicles (AUSVs) represent a distinct class of automated platforms designed to operate at very low speeds (often around 5–10 km/h) for urban road maintenance tasks, yet their integration in mixed traffic remains poorly understood. This study evaluates the effect of AUSVs through a microscopic simulation using a real-world network modeled on the Pangyo autonomous driving testbed in South Korea. Twelve scenarios (six AUSV speeds, 5–30 km/h, and two traffic volumes, level of service [LOS] C and LOS D) were simulated; LOS C and D were deliberately selected—beyond the typical off-peak LOS A/B context—to empirically identify the demand threshold above which AUSV effects become operationally significant. Performance was assessed across vehicle, lane, link, and network levels using speed, shockwave propagation, time-to-collision (TTC) ratios, and lane-change rates. Results revealed consistent patterns. At the lane level, AUSVs at 5 km/h induced backward-moving queues up to 120 m in length and critical TTC ratios of up to 0.50 under LOS D, while no significant disturbances occurred under LOS C—revealing a critical operational threshold between the two demand regimes. At the link level, speed reductions of ≥20 km/h concentrated on two-lane bottlenecks, where overtaking opportunities were constrained. Network-wide, average speeds increased monotonically with AUSV speed, while lane-change rates peaked at 5 km/h (32.5 events/km under LOS C; 59.1 events/km under LOS D) and declined thereafter. These findings support demand-aware scheduling (off-peak for ≤10 km/h operation), minimum speed thresholds (≈15 km/h), and bottleneck-avoiding routes as practical strategies for sustainable AUSV deployment.

Preparing the Transportation Workforce for Emerging Technologies: A Guide

Sat, 16 May 2026 12:15:37 GMT

This report presents strategies with tailored resources to support efforts by transportation agencies to recruit, develop, and retain staff capable of deploying new transportation technologies. The guide was developed on the basis of an extensive review of industry workforce needs and challenges found in the literature, in stakeholder engagement, and practitioner workshops. This guide will be valuable to transportation agency staff and leadership seeking to build internal capacity to adopt and leverage the benefits of emerging transportation technologies.

Mechanisms to Address Off-Airport Obstructions

Sat, 16 May 2026 12:15:37 GMT

This report presents the state of practice of airport methods to address obstructions located off airport property. The synthesis includes information on activities airports take to address obstructions, including outreach with landowners and other stakeholders, time and costs to resolve issues, and support and coordination from local, state, and federal authorities. Under ACRP Project 11-03/Topic S09-11, “Survey of Mechanisms to Address Off Airport Obstructions,” Embry-Riddle Aeronautical University was asked to synthesize and document the various mechanisms airports use to address obstructions, with a focus on obstructions that are outside of the airport boundary. Information used in this study was obtained through a literature review, a survey of airports, and interviews to develop in-depth case examples. This synthesis is an immediately useful document that records the practices that were acceptable within the limitations of the knowledge available at the time of its preparation. As progress in research and practices continues, new knowledge will be added to that now at hand. The audience for this synthesis is airport sponsors, local permitting authorities, state aviation officials, and non-airport stakeholders that are involved in addressing off-airport obstructions.

Addressing Liability Issues of Proactive Safety Improvements

Sat, 16 May 2026 12:15:36 GMT

This report examines the legal considerations facing transportation agencies that adopt proactive, data-driven approaches to roadway safety. The research analyzes how predictive safety methods, used to identify roadway features and locations associated with elevated crash risk, may intersect with tort liability concerns. The findings are intended to assist state departments of transportation in understanding legal risks, available defenses, and strategies to manage liability while advancing safety objectives. The digest will be of particular interest to agency counsel, risk managers, and senior transportation officials. Transportation agencies increasingly rely on proactive safety analysis to inform roadway design and improvement decisions before crashes occur. However, uncertainty regarding the potential use of proactive safety methodologies, manuals, and guidance in tort litigation has raised concerns about increased exposure to liability. Although federal law limits the admissibility of certain safety data and studies, those protections do not clearly extend to safety manuals or predictive analytical tools. Research was needed to assess whether and how tort liability concerns may affect the adoption of proactive safety approaches. Under NCHRP Project 20-06/Topic 27-05, the Fine Points, Ltd. research team was asked to examine the legal landscape surrounding proactive safety improvements. This study reviews relevant statutes, case law, and agency practices; identifies litigation in which proactive safety concepts have been raised; evaluates defenses and outcomes; and documents strategies used by transportation agencies to mitigate liability concerns. Based on this analysis, the research explores potential legal, administrative, and policy approaches that may support the use of proactive safety methods while preserving established processes for prioritizing and funding roadway improvements.

Integrated Container Slot Allocation and Automated Stacking Crane Scheduling in Automated Container Terminals with Limited Buffers

Sat, 16 May 2026 12:15:36 GMT

This paper investigates the integrated scheduling of automated stacking cranes (ASCs) and container slot allocation in automated container terminals (ACTs) with limited buffer capacity. A hybrid stacking strategy based on time windows is proposed, and a bi-objective mixed-integer programming model is developed, considering automated guided vehicles and truck buffer capacities, ASC safety distances, and handshake area operations. An enhanced non-dominated sorting genetic algorithm II with tabu search (NSGA-II-TS) is designed, with parameters optimized via sensitivity analysis. Experiment results show that comparisons with an exact mixed-integer linear programming solver validate the solution quality of the proposed approach, and that the proposed algorithm significantly outperforms benchmark heuristic methods in generating high-quality Pareto-optimal solutions. Case studies reveal that dynamically adjusting handshake area locations and setting buffer capacity to six units effectively balance container flow and operational costs. The proposed approach is also validated against two alternative scheduling strategies, demonstrating superior effectiveness. This research provides new strategies and a robust method for improving the operational efficiency of ACTs under buffer constraints.

Experimental Study on the Impact Force and Dynamic Evolution of Landslide-Debris Flows at the Portal Section of Mountain Tunnels

Sat, 16 May 2026 12:15:36 GMT

As transport networks continue to improve, more tunnels are being commissioned. The complex topography and geological conditions of certain mountainous regions have led to the construction of road and railway tunnels that must pass through geologically sensitive zones which are prone to landslide-debris flows at tunnel entrances during operation. This study, based on real engineering cases, combines scale-model tests and numerical simulations to examine the maximum impact force and dynamic behavior of landslide-debris flow at tunnel entrances in mountainous areas. The main findings are as follows. (1) Laboratory model tests were used to quantitatively analyze how particle gradation, flume inclination, and source volume influence impact force, and particle gradation was identified as the most significant factor. (2) A three-dimensional physical model of landslide-debris flow was developed, based on the physical and mechanical properties of landslide-debris-flow particles and their size distribution obtained from testing. This model simulated the entire process, from the initiation of landslide-debris flows to their attenuation, revealing the evolution of flow depth and velocity during impact. (3) The debris flow accumulates from east to west, reaching a maximum depth of about 12 m at the tunnel entrance, posing a considerable threat. The impact process can be divided into three stages: 0–6 s of initial acceleration, 6–13 s of slow acceleration, and 13–23 s of deceleration and cessation of accumulation. Because particle size substantially influences the impact behavior of debris flows, we recommend adopting a tiered protection system that classifies and designs countermeasures according to particle-size composition.