Traffic flow prediction is an important component of intelligent transportation systems. Recently, unprecedented data availability and rapid development of machine learning techniques have led to tremendous progress in this field. This article first introduces the research on traffic flow prediction and the challenges it currently faces. It then proposes a classification method for literature, discussing and analyzing existing research on using machine learning methods to address traffic flow prediction from the perspectives of the prediction preparation process and the construction of prediction models. The article also summarizes innovative modules in these models. Finally, we provide improvement strategies for current baseline models and discuss the challenges and research directions in the field of traffic flow prediction in the future.
Taxi demand prediction is a crucial component of intelligent transportation system research. Compared to region-based demand prediction, origin-destination (OD) demand prediction has a wide range of potential applications, including real-time matching, idle vehicle allocation, ride-sharing services, and dynamic pricing, among others. However, because OD demand involves complex spatiotemporal dependence, research in this area has been limited thus far. In this paper, we first review existing research from four perspectives: topology construction, temporal and spatial feature processing, and other relevant factors. We then elaborate on the advantages and limitations of OD prediction methods based on deep learning architecture theory. Next, we discuss ongoing challenges in OD prediction, such as dynamics, spatiotemporal dependence, semantic differentiation, time window selection, and data sparsity problems, and summarize and compare potential solutions to each challenge. These findings offer valuable insights for model selection in OD demand prediction. Finally, we provide public datasets and open-source code, along with suggestions for future research directions.
After some tragic fire events, Directive 2004/54/EC was issued to ensure a minimum safety level for tunnels belonging to the Trans-European Road Network longer than 500 m. Nowadays, most of the Italian road tunnels are still not in compliance with the minimum safety requirements, thus refurbishment works are often planned. By developing a traffic macro-simulation model, this paper aims at assessing the resilience of an existing twin-tube motorway tunnel when one of its tubes is partially or completely closed due to planned activities. Several scenarios were investigated, also considering the availability or not of an alternative itinerary in the surrounding transportation network. The average vehicles’ speed was used as a functionality parameter, while the resilience metrics were the resilience loss, the recovery speed, and the resilience index. The findings showed higher resilience losses for complete closure rather than partial closure of the tube under planned refurbishment works. The implementation of digital technologies, such as variable message signs, might reduce the resilience loss of the tunnel system. This research might represent a reference for tunnel management agencies in the choice of the most appropriate traffic control strategy to improve tunnel resilience in the event of planned activities.
Countdown signals for motorized vehicles, which are intended to ensure safety on the road and regulate motor vehicle speed limits at road intersections, are still considered a relatively novel concept. These signals have been adopted by only a few countries, and the number of cities that use them is limited. This review aims to summarize the effects of countdown signals on traffic safety and efficiency and to determine the consistency and differences of existing research propositions on the matter. Based on the review, considerable research presents evidently different conclusions in the areas of driver red-light running and traffic safety. Particularly, some studies propose that countdown signals reinforce traffic safety, whereas others consider that such signals adversely affect traffic safety. Meanwhile, related literature provides varying conclusions on the aspect of traffic efficiency for vehicle headway. At present, the number of studies conducted regarding the driving behaviors of motorists toward countdown-signalized intersections is insufficient. Accordingly, such inadequate diversity in research causes difficulty in completely assessing the benefits and disadvantages of countdown signals. In this paper, an important future research direction on microcosmic driving psychological and physiological data combined with macro-driving behavior is proposed.
Connected vehicle (CV) is regarded as a typical feature of the future road transportation system. One core benefit of promoting CV is to improve traffic safety, and to achieve that, accurate driving risk assessment under Vehicle-to-Vehicle (V2V) communications is critical. There are two main differences concluded by comparing driving risk assessment under the CV environment with traditional ones: (1) the CV environment provides high-resolution and multi-dimensional data, e.g., vehicle trajectory data, (2) Rare existing studies can comprehensively address the heterogeneity of the vehicle operating environment, e.g., the multiple interacting objects and the time-series variability. Hence, this study proposes a driving risk assessment framework under the CV environment. Specifically, first, a set of time-series top views was proposed to describe the CV environment data, expressing the detailed information on the vehicles surrounding the subject vehicle. Then, a hybrid CNN-LSTM model was established with the CNN component extracting the spatial interaction with multiple interacting vehicles and the LSTM component solving the time-series variability of the driving environment. It is proved that this model can reach an AUC of 0.997, outperforming the existing machine learning algorithms. This study contributes to the improvement of driving risk assessment under the CV environment.
The exhaust emissions and frequent traffic incidents caused by traffic congestion have affected the operation and development of urban transport systems. Monitoring and accurately forecasting urban traffic operation is a critical task to formulate pertinent strategies to alleviate traffic congestion. Compared with traditional short-time traffic prediction, this study proposes a machine learning algorithm-based traffic forecasting model for daily-level peak hour traffic operation status prediction by using abundant historical data of urban traffic performance index (TPI). The study also constructed a multi-dimensional influencing factor set to further investigate the relationship between different factors on the quality of road network operation, including day of week, time period, public holiday, car usage restriction policy, special events, etc. Based on long-term historical TPI data, this research proposed a daily dimensional road network TPI prediction model by using an extreme gradient boosting algorithm (XGBoost). The model validation results show that the model prediction accuracy can reach higher than 90%. Compared with other prediction models, including Bayesian Ridge, Linear Regression, ElatsicNet, SVR, the XGBoost model has a better performance, and proves its superiority in large high-dimensional data sets. The daily dimensional prediction model proposed in this paper has an important application value for predicting traffic status and improving the operation quality of urban road networks.
Implementing autonomous bus services in several cities has garnered substantial research attention worldwide. However, the benefits and challenges of this emerging mode remain insufficiently understood. Consequently, VOSviewer was employed for a bibliometric analysis involving 300 publications, investigating the associations among authors, journals, and keywords. Subsequently, we comprehensively reviewed the current state of research on two topics and proposed future recommendations. Results indicate that the first document related to autonomous bus services was published in 2009. Most user attitude -related research data are obtained via questionnaires and analyzed using statistical techniques. Autonomous bus services are expected to benefit passengers regarding travel time, cost, safety, etc., while passenger preferences are inconsistent. However, integrating the service into existing bus systems requires careful consideration of the schedule sequences. Notably, modular autonomous bus services present a new opportunity for the further optimization of bus services. In future studies, standardized data acquisition procedures should be developed to achieve comparable results. Regarding traveler choice behavior, the effect of specific autonomous bus service policies over time and the heterogeneity due to cultural or social contexts across regions should be assessed. To further promote autonomous bus services, based on fluctuating travel demands, the effects of vehicle capacity, speed, and cost of fleet composition should be evaluated comprehensively to optimize the bus network and schedule sequence. Owing to the protracted nature of the transition from conventional to fully autonomous buses, one should prioritize semi-autonomous bus services. Another essential future research direction is to integrate modular autonomous bus assembly or disassembly strategies with different fine-grained operation optimization techniques in various scenarios.
