Journal of Transportation Engineering, Part A: Systems

When responding to changes in the status of traffic controls, drivers may experience uncertainties caused by incomplete and inaccurate information. These uncertainties may result in driver anxiety during the decision-making process and lead to poor decisions. It is hypothesized that a better understanding of the decision-making process, and the associated gate violation behavior at highway-rail grade crossings (HRGCs), will lead to effective gate violation countermeasures and, ultimately, safer HRGCs. This paper examines the relationship between driver anxiety and the characteristics of inappropriate decisions that result in gate violations. Field data were collected at a gated HRGC test bed in Lincoln, Nebraska. A total of 372 vehicles were studied which included 116 first-to-stop and 256 last-to-proceed vehicles, with 76 proceeding vehicles identified as gate violations. A fuzzy-based, degree of anxiety metric is adopted as a surrogate measure of driver anxiety in the decision-making process. The results show that HRGC gate violations were linearly related to the anxiety metric values. In other words, all things being equal, as the anxiety value of the metric increased so did the probability of a violation. The maximum anxiety was felt when drivers were approximately 5 s from the stop line at the onset of the HRGC’s flashing lights. The results provide guidance for applying ancillary or advanced technologies for increasing HRGC safety, such as where to install a roadside prepare-to-stop system, or when to issue an in-vehicle warning in the context of automated driving systems.

One fundamental issue in managing a shared-parking system is predicting shared-parking demand. Such predictions are very challenging because predicting shared-parking demand usually involves nonlinearities and complex spatiotemporal dependencies. In this paper, we propose a deep learning–based network comprising of three modeling components—CNN-Module, Conv-LSTM-Module, and LSTM-Module—to predict the shared-parking inflow and outflow in each region of a shared-parking system. First, the CNN-Module utilized convolution neural networks to determine local spatial dependencies. Second, the Conv-LSTM-Module leveraged the Conv-LSTM neural network to capture similarities of shared-parking demand across different regions. Finally, the LSTM-Module was applied to model temporal features by using the Long Short-Term Memory (LSTM) network. Moreover, we also divided the input into three components (recent, daily, and weekly) to extract the periodically shifted relations. The model was evaluated using a real-world shared-parking data set in Chengdu, China. Experiments showed that our model outperforms six other well-known baseline methods within an acceptable time frame. Extensive additional experiments and evaluations were conducted to investigate the sensitivity of our model.

Examining the relationship between congestion and travel time reliability (TTR) can help in gaining a broader perspective on the development of congestion mitigation strategies. Several performance measures have been proposed and used by researchers and practitioners to assess congestion levels on a freeway in the past. These performance measures identify congestion when the space mean speed of a segment is below a critical static threshold and does not vary with changes in road geometry, land use, area type, and presence of acceleration or deceleration lane. This research integrates travel time data and demonstrates the applicability of a probabilistic approach based on hazard function to quantify congestion, characterize the congestion levels, and generate reliability-based level-of-service (LOS) thresholds for freeways. Toward this purpose, travel time data for 44 geographically distributed freeway segments in an urban area with varying speed limits and road characteristics were used to derive the travel time threshold ($T_c$) representing the lower boundary limit of congested condition using a probabilistic approach. A new measure called congestion index (CI) was then proposed to assist in quantifying both the intensity and duration of congestion. The CI accounted for the variation in road characteristics, spatial location, and traffic condition. Further, the CI was observed to be correlated between continuous segments, highlighting that the CI could characterize recurring congestion. The CI was correlated with the TTR measures to generate reliability-based LOS thresholds for freeways. A strong positive correlation (correlation coefficient greater than 0.80) was observed between the congestion level and TTR measures. This research demonstrates the applicability of the reliability-based LOS for prioritizing segments and effectively allocating funds. The generated reliability-based LOS thresholds can help practitioners assess the performance of freeways in terms of CI and TTR, which can be followed up with microlevel analysis to identify congestion mitigation strategies.

A two-stage method for matching private idle parking slots with demanders is proposed based on the core idea of private idle parking slot sharing to solve the problem of parking difficulty in a shared economy environment. This method combines knowledge rules and a multiobjective optimization model. In the first stage, knowledge rules for screening private idle parking slots that meet demanders’ walking distance and parking price requirements are proposed. In the second stage, the degree of satisfaction of demanders concerning parking price and walking distance from private idle parking slots to their destination and the degree of satisfaction of private idle parking slots regarding parking slot occupancy rates are constructed first. Then, considering whether private idle parking slot owners provide extended time, a multiobjective optimization model is established by maximizing the satisfaction degree of demanders, the satisfaction degree of private idle parking slot owners, and the profits of the platform. Further, an improved nondominated sorting genetic algorithm II (INSGA II) is developed to solve the model. Finally, a practical example is used to illustrate the feasibility and effectiveness of the proposed method.

This study aims to build an accurate merging position model by incorporating heterogeneity into the accelerated hazard model based on random parameter and latent class models. Three kinds of distribution (Weibull, log-normal, and log-logistic) are compared. The Weibull distribution is found to outperform the others. The latent class accelerated hazard model better captures the unobserved heterogeneity among drivers. The data used in this study can be naturally segmented into two classes. The large class consists of about 80% of drivers and the estimation results are similar to fixed parameter and random parameter models. The smaller one contains drivers that only consider the subject vehicle’s speed; these drivers merge rather late. Some of these drivers even merge into the adjacent main lane using the shoulder lane when they cannot find suitable gaps at the end of the auxiliary lane. Considering its good performance, the proposed latent class accelerated hazard model could also bring new insights into road design, such as the determination of the length of the acceleration lane or auxiliary lane.

This paper develops a diagnostic-oriented and evaluation-driven framework that can bridge the gap between transit performance evaluation and improvement prioritization at the route level, where the roles of government agencies, transit operators, and passengers are incorporated. In the proposed framework, input and output variables are extracted from multiple data sources to evaluate technical efficiency and service effectiveness of bus routes, and after which, the inefficient bus routes are analyzed through an improvement diagnosing process featured with route categorization and improvement direction identification. Three alternative strategies, namely, adjusting single critical input or output strategy, adjusting multiple critical input or output strategy, and simultaneously controlling both input and output strategy are developed. Recommendations for selecting these strategies are also provided for transit agencies to develop practical solutions. The framework is applied and validated in a case study for the service performance improvement of 23 bus routes in Jinan, China.

This work supported the Federal Highway Administration (FHWA) in presenting state and metropolitan area vehicle occupancy information in compliance with Title 23 of the US Code of Federal Regulations, Part 490 National Performance Measures. The basic goal is to provide and introduce a statistically effective and realistic approach to approximate bus occupancy rates for each US state and Washington, DC. Bus occupancies were calculated separately for transit buses, school buses, and motorcoaches. The average total bus occupancy was determined by aggregating the average occupancy of the vehicle for the three groups weighted by annual vehicle miles traveled. Regarding transit buses, the National Transit Database (NTD) of the Federal Transit Administration (FTA) was used to measure the occupancy of transit buses. For school buses, state transportation statistics from 2015–2016 were used to measure average school bus occupancy for each state. Data provided by the Port Authority of New York and New Jersey (PANYNJ) for the Port Authority Bus Terminal (PABT) in New York City, the largest US bus terminal, were used to measure the occupancy rates for motorcoaches. Results on state-level bus occupancy rates are summarized. The paper concludes with guidelines for future data collection, validation, and training. The date, code, and user guide of this study can be found at an online GitHub repository.