Journal of Transportation Engineering, Part A: Systems

The past few years have been witness to an increase in autonomous vehicle (AV) development and testing. However, even with a fully developed and comprehensively tested AV technology, AVs are anticipated to share the roadway network with human drivers for the unforeseeable future. In such a mixed environment, we use naturalistic driving data from the next generation simulation (NGSIM) and Lyft Level 5 (Lyft L5) prediction data sets to investigate whether the existence of AVs influences the car-following behavior of human drivers. We use time headway time series as a proxy to capture the car-following behavior of human drivers. We then develop a nested fixed model to find possible changes in behavior when human drivers are following different types of vehicles (i.e., human-driven vehicles or AVs). The factors included in this model are the platoon structure (a legacy vehicle following a legacy vehicle, and a legacy vehicle following an autonomous vehicle), road type (freeway and urban), time period (morning and afternoon), lane (right, middle, and left), and the source of the data (NGSIM and Lyft L5). Results indicate a statistically significant difference between the car-following behavior of drivers when they follow a human-driven vehicle compared to an AV. This change in the car-following behavior of drivers is manifested in the form of a reduction in the mean and variance of time headways when human drivers follow an AV. These findings can bridge the gap between anticipated and real-world impacts of AVs on traffic streams and roadway capacity, providing meaningful insights on the influence of AVs on the driving behavior of humans in a naturalistic driving environment.

The fatal or incapacitating injury caused by overturn crashes involving sport utility vehicles (SUVs) is irreparable. The purpose of this study is to identify potential factors that affect the injury severity of overturn crashes involving SUVs and develop adequate preventive strategies. Given the unobserved heterogeneity existing in the data set, crash data in North Carolina from Highway Safety Information System (HSIS) is analyzed and separated by Latent Class Clustering into six relatively homogeneous groups. To further explore the heterogeneity, random parameter logit models are developed for each cluster, and the impacts of significant factors are estimated with marginal effects. The results reveal the heterogeneity across the clusters and the homogeneity within the same cluster. Variables (including females, people over fifty years old, improper or aggressive behavior, rural areas, high-speed limit, curved roadway, rolling and mountainous terrain, adverse weather, and poor light conditions) are associated with the injury severity of the overturn crashes involving SUVs. The findings of this study can further provide decision makers with insightful countermeasures to improve transportation safety and mitigate the injuries of overturn crashes involving SUVs.

Left-turn execution in autonomous driving at urban intersections is often complex and characterized by unpredicted events, such as vehicles speeding and running a red light. Despite these hazards, autonomous vehicles must drive through intersections safely and efficiently. To solve this problem, a new hierarchical driving strategy (HDS) is proposed for connected and autonomous vehicles making a protected left turn at signalized intersections, which combines the rule-based method and deep reinforcement learning (DRL). The high level of the HDS is the rule-based decision-making module, and the low level is the driving skill, which is dependent on the status. Specifically, the vehicle status when turning left is divided into safe and alert statuses. Further, DRL is used to train the driving skill of the vehicle for each status. The HDS design effectively combines the advantages of end-to-end driving. Moreover, the algorithm was experimentally evaluated in multiple protected left-turn scenarios. Compared with the pure rule-based method, the HDS achieved a 34% reduction in failure rate in the test scenario, and the driving behavior of the autonomous vehicle employing HDS was more intelligent. Moreover, the HDS is highly robust to complex scenarios.

Different original equipment manufacturers (OEMs) provide different connected and autonomous vehicle (CAV) solutions. However, previous studies that estimated the impact of CAV on driving behaviors assumed CAVs are identical, with an integrated form of technologies. In this study, a simulation was conducted to investigate how different fundamental elements of CAV technology (i.e., communication range, number of vehicular interactions, and car-following distance) affect right-turn gap acceptance behavior at unsignalized intersections. The results of the statistical analysis, critical gaps comparison, platooning analysis, and logistic regression revealed that (1) all elements increase the lag acceptance rate compared with all human-driven vehicles; (2) a longer communication range and more vehicular interactions make CAVs turn conservatively from a minor road, causing the critical gap to increase; (3) a shorter car-following distance causes CAVs to generate longer platoons and shorter gaps on a major road, which decreases acceptance probability; and (4) CAVs with a shorter car-following distance on a minor road are more aggressive in accepting gap/lag, causing the critical gap to decrease and acceptance probability to increase. The study conducted an impact analysis of CAV technologies from a novel perspective, and the results can be used as a technical report of CAV’s performance to help build transportation policies.

Traffic breakdowns are frequently observed phenomena on roads in larger cities, especially during peak hours. Locations along a road stretch with frequently observed breakdowns are known as recurrent bottlenecks. Knowledge about bottleneck locations are important for improvement of traffic conditions at these locations. Bottleneck locations can be identified through manual inspection of data. However, due to the comprehensive amount of data that are available today, it becomes impractical to manually identify breakdowns and instead, an automated process can be used. We propose such an automated method. The proposed method is applied to a use case south of Stockholm in Sweden. One month of data collected at densely spaced detectors is used to investigate the sensitivity of the parameter settings. After calibration of the threshold values, 100% of the larger breakdowns and 40% of the medium size breakdowns are identified. Smaller breakdowns, not giving significant impact on the traffic conditions, are only detected in 10%–20% of the cases. Thereafter, the method is applied to 1 year of data to illustrate the applicability of the method on a larger data set. The results show that the method is promising to use for identification of recurrent bottleneck locations.

The current paper implements a methodology for automatically detecting vehicle maneuvers from vehicle telemetry data under naturalistic driving settings. Previous approaches have treated vehicle maneuver detection as a classification problem, although both time series segmentation and classification are required since input telemetry data are continuous. Our objective is to develop an end-to-end pipeline for the frame-by-frame annotation of naturalistic driving studies videos into various driving events including stop and lane-keeping events, lane changes, left-right turning movements, and horizontal curve maneuvers. To address the time series segmentation problem, the study developed an energy-maximization algorithm (EMA) capable of extracting driving events of varying durations and frequencies from continuous signal data. To reduce overfitting and false alarm rates, heuristic algorithms were used to classify events with highly variable patterns such as stops and lane-keeping. To classify segmented driving events, four machine-learning models were implemented, and their accuracy and transferability were assessed over multiple data sources. The duration of events extracted by EMA was comparable to actual events, with accuracies ranging from 59.30% (left lane change) to 85.60% (lane-keeping). Additionally, the overall accuracy of the 1D-convolutional neural network model was 98.99%, followed by the long-short-term-memory model at 97.75%, then the random forest model at 97.71%, and the support vector machine model at 97.65%. These model accuracies were consistent across different data sources. The study concludes that implementing a segmentation-classification pipeline significantly improves both the accuracy of driver maneuver detection and the transferability of shallow and deep ML models across diverse datasets.

The acceptance of railway systems as a frontier transportation infrastructure can be attributed to their reliability, safety, and support for green technology. With the recent advances in artificial intelligence and machine learning (AI/ML), the maintenance of railroad transportation systems has taken a different direction, especially in the analysis of railroad big data, leading to real-time processing and detection of railway problems. However, using limited track data may result in overfitting, hindering the accurate implementation of robust models. In this paper, the authors consider generative adversarial networks (GANs) with keen consideration for possible covariate shifts to improve track defect detection and decrease data imbalance. The results show that implementing covariate-shift GAN (COGAN) reduces image processing time and eliminates image biases.

Variable speed limit (VSL) and variable advisory speed (VAS) signs are efficient, cost-effective and among the state-of-the-art active traffic management (ATM) strategies. They adopt the idea of dynamically changing posted speed limits to improve highway safety performance and operation by harmonizing traffic speed. VSL/VAS system involves changing speed limits according to real-time traffic events and weather conditions. Hence, traditional average annual daily traffic (AADT) based crash prediction models may not capture the temporal effect of traffic characteristics due to the high level of aggregation. To address this issue, short-term safety performance functions (SPFs) with aggregation levels of average annual weekday hourly traffic (AAWDHT) and average annual weekday peak traffic (AAWDPT) along with AADT-based SPFs were developed using high-resolution traffic detector and VSL/VAS operational data. In this study, the Poisson log-normal model performed well at each level of aggregation and thus is recommended for developing short-term SPFs. In line with previous studies, traffic volume and standard deviation of speed were found to be positively associated with crash frequency in all the estimated models. In addition, it was found that implementation of VSL/VAS significantly reduced crash frequency by 15.97% and 26.14% for the AAWDHT and AAWDPT models, respectively. The safety improvement was captured in the short-term models in a more distinguished way than in the highly aggregated AADT-based model. The findings of this study pave the way for practitioners and policymakers to evaluate and select important parameters for VSL/VAS strategy implementation on freeways.

Work zones pose mobility issues to the traveling public and safety challenges to travelers and road maintenance workers. These safety and mobility issues may be exacerbated by the presence of heavy vehicles. Connected and automated vehicle (CAV) technologies have been identified as a potential solution for these issues. This paper investigates the operational impacts of connected and automated heavy vehicles (CAHV) on freeway work zone operations on interstate highways. A microsimulation model, calibrated to empirical work zone field data, was used to study the operational impacts of CAHV platoons under various work zone and traffic conditions. It was found that, as the CAHV market penetration rate increases, the average work zone delay and queue length decreases. In addition, as the demand and heavy vehicle percentage increases, so do the benefits of using CAHV technology. For example, it was found that, when all heavy vehicles are classified as CAHV, the average flow rate is approximately 67% higher, and the maximum queue size and average delay decrease by approximately 97%. The methodology used in this paper will help transportation agencies as they design work zones to accommodate heavy vehicles equipped with CAV technologies.