Eastern European Economics

AbstractThe development of the civil aviation industry has contributed to a steady increase in the number of daily flight operations at airports, which in turn has led to increasingly complex airport ground layouts. To aid airport managers in understanding the operational situation on the airport surface, this paper introduces a predictive model for airport ground conflict situations based on GNN‐LSTM. This model identifies potential conflicts, conflict hotspots, and conflict hotspots zones, designating key intersections on taxiways as conflict hotspots according to taxiing rules. A conflict network is constructed, employing GNN with an integrated attention mechanism to extract structural features of the network, while LSTM is utilized to capture temporal features. After tuning the model parameters, predictions are made regarding the overall potential number of potential conflicts on the surface. To validate the effectiveness of the model, experimental analysis is conducted using AirTOp simulation data from Shenzhen Bao'an Airport, comparing GNN‐LSTM model with GNN‐GRU, LSTM, and GRU models, using RMSE and MAE as loss functions. The results demonstrate that he proposed modelling approach effectively extracts the temporal features of potential conflict and GNN‐LSTM model outperforms other models in predicting the overall number of potential conflicts.

AbstractBoarding stop inference for bus passengers is essential for the improvement of bus transit services. Previous studies mainly focus on matching the bus trajectories with the bus stop locations, while the relationship between smart card readers—which collect the smart card data—and bus vehicles is usually given. However, uncertainties arise in practical applications regarding the matching of vehicles and card readers. To tackle this challenge, in this study, a data‐driven approach is proposed to dig into the spatiotemporal features of passengers' smart card data and bus vehicle operations. A weighted bipartite graph algorithm is developed to match the smart card readers with the bus vehicles automatically. To verify the feasibility and effectiveness of the proposed approach, a case study is conducted on the Bus Anhong Line in Shanghai, China. The inferred results of boarding stops are compared with the data from passenger counting sensors installed in the bus vehicles. The matching accuracy rate achieves 0.9539, which validates the effectiveness of the proposed matching model. In addition, the inferred data are used to present the spatiotemporal patterns of boarding passengers and identify high‐demand bus stops.

AbstractThis case study introduces an innovative multivariate methodology for assessing the lifetime of marine engineering systems, specifically in cargo vessel transportation. The analysis focused on stress data collected onboard a 4400 TEU container vessel during multiple trans‐Atlantic voyages. One of the major challenges in marine cargo transport lies in mitigating the risk of container loss due to excessive whipping loads. Accurate prediction of extreme stress levels on vessel deck panels remains difficult, primarily because of the nonlinear and non‐stationary nature of wave and ship motion interactions. Higher‐order dynamic effects, such as second‐ and third‐order responses, often become significant when ships operate under adverse environmental conditions, amplifying nonlinear influences. Laboratory simulations, constrained by wave characteristics and scale similarity issues, may not always provide reliable results. Consequently, data collected from vessels navigating extreme weather conditions serves as a critical resource for comprehensive container ship risk assessment. The primary goal of this study was to validate and demonstrate the effectiveness of a novel multivariate risk evaluation approach, leveraging onboard measurements of dynamic areal pressure on cargo ship deck panels as the core dataset. The Gaidai methodology for multivariate risk evaluation proved to be a robust tool for assessing failure, hazard, and damage risks in complex, nonlinear vessel deck panel and ship hull stress systems.

AbstractThis work presents the design and implementation of a reinforcement learning‐based autonomous parking system where an agent is trained to reverse‐park in a selected parking spot. The parking procedure is divided into three stages, and each stage has its corresponding surrogate objective that contributes to the overall parking process. The model solely depends on features extracted from a top‐view image of the parking space. It has the advantage of potential deployment in smart parking buildings without refitting non‐autonomous cars with modern sensors. The training was conducted offline on a simulation utilizing the proximal policy optimization algorithm. The model was then transferred and tested on a hardware prototype of the parking space. The results of the system were successful as the successful parking rate reached 100% with no collisions with any objects, and the fastest parking time reached 10 s. The testing was conducted on multiple samples and scenarios of the parking setup.

AbstractCooperation classes have been defined by SAE International to differentiate the communication capabilities between vehicles and infrastructure. To advance understanding of the impact of cooperation classes on autonomous cooperative driving and optimize traffic operations, this article proposes an edge‐computing‐based operations framework for cooperative‐automated driving system (C‐ADS)‐equipped vehicles at a stop‐controlled intersection. First, a critical time points estimation component estimates a set of critical time points for each C‐ADS‐equipped vehicle. Second, a trajectory‐smoothing component is called at each C‐ADS‐equipped vehicle in a decentralized manner to control C‐ADS‐equipped vehicle trajectories based on the estimated critical time points and its cooperation behavior. Notably, this study represents a first‐time investigation of different cooperation classes for stop‐controlled intersections. Simulation results show that the proposed framework can reduce stop‐and‐go traffic, yielding significant improvements in mobility and energy efficiency, as the cooperation class increases. Results also demonstrate that the proposed framework is suitable for real‐time applications by distributing computational burden in different entities. Further, results verify that the proposed framework can handle varying speed errors without significant loss in performance.

AbstractRecent transport policies increasingly promote shifts towards rail travel aiming at a more sustainable transportation system. This shift is hampered by widespread unexpected perturbations in operations, resulting in perceived poor punctuality and reliability. When prevention of such perturbations is not feasible, traffic management must mitigate their effects, resolving arising conflicts to restore regular train operations and minimize delay. Current practice generally includes the assessment of railway performance in terms of train delays, but the quality of service to passengers is rarely explicitly accounted for. A railway traffic management framework is proposed that accounts for both passenger and train delays. To do so, a predictive optimization framework is proposed, integrating a demand prediction module, a passenger demand assignment module and a traffic management module. The first dynamically predicts future origin‐destination passenger flows using linear regression on real‐time observed smart card data. Then, the demand assignment module links predicted passengers to specific train paths, given a railway schedule. Finally, the traffic management module optimizes train scheduling and routing in real time, under the combined objective of minimizing train and passenger delays. The methodology is validated and benchmarked against equivalent passenger agnostic traffic management on a case study of the Copenhagen suburban railway network. The results show that it is possible to take into account passenger perspective in railway traffic management, without reducing the railway system efficiency compared to classic approaches.

AbstractAnalysing post‐accident scenes using in‐vehicle cameras is crucial for effective highway traffic control and enhancing accident response, road safety, and traffic flow. This contributes to a comprehensive understanding of the situation and achieves better decision‐making and effective management. The accident scene report system is designed to focus on specific post‐accident objects, such as crashed vehicles, involved individuals, emergency vehicles, and debris. This means that the post‐accident object detection algorithm needs to handle a wide variety of objects, from large collapsed vehicles to tiny particles. It should operate in real‐time on embedded boards, balancing detection accuracy and compactness to fit within the constraints of embedded computing modules. This approach aims to facilitate prompt reporting to traffic control centres. In this study, a hybrid spatial and channel attention and its pruning algorithm tailored for object detection in post‐accident scenarios are proposed. This approach markedly enhances the detection performance in the unexpected accidents and malfunctioning scenes, significantly boosting the system's accuracy and processing speed. The method optimally balances the model compactness with seamless attention and pruning, making it highly suitable for real‐time applications in traffic monitoring systems. The proposed seamless attention and pruning method is demonstrated using the proposed accident object detection dataset.

AbstractIntelligent vehicles are prone to dangerous issues such as sideslip and instability when changing lanes to avoid obstacles under some extreme conditions. Therefore, to improve safety and stability during the obstacle‐avoidance process, this paper proposes a lane‐change control method that considers yaw stability based on distributed drive electric vehicles. Fuzzy adaptive model predictive control and proportional integral derivative (PID) control are, respectively, established to compute the optimal front wheel steering angle and vehicle longitudinal torque under lateral and longitudinal decoupling. Additionally, a direct yaw moment controller is constructed based on model predictive control to calculate the additional yaw moment, which is then distributed according to the tire adhesion utilisation rate to optimise yaw stability in lane‐changing obstacle‐avoidance scenarios. Finally, the proposed control framework is verified in typical obstacle‐avoidance scenarios. The results show that, compared to the control method that do not consider yaw stability, the average yaw rate deviation is reduced by 54.0% on high‐adhesion road surfaces and by 61.2% on low‐adhesion road surfaces, achieving further optimsation in the safety and stability of the obstacle‐avoidance process.

AbstractThis paper proposes a two‐layer control strategy to enhance connected vehicles fuel efficiency. Compared with previous works, this study proposes an approach to optimize both the vehicle speed and gearbox position to achieve better fuel efficiency. The control task is given in two stages: the upper layer and the lower layer. Before trip departure, the upper layer concurrently optimizes the vehicle speed and gearbox position based on road map information, and engine and vehicle parameters for an entire route. Then, while driving, the lower layer is used to follow the pre‐computed optimal profiles. Model predictive control follows the optimal speed while ensuring an adaptive safe distance constraint with a preceding vehicle. For gear shifting, an online shift control assuring the tracking of the optimal gear position is developed based on look‐ahead road data and vehicle actions. The effectiveness of the proposed control strategy was evaluated with comprehensive simulations and comparison tests using Matlab and CarSim software. The mean online optimization calculation time is 0.065 s, indicating its real‐time capability. The proposed method can be used as a driving assist system or implemented as a speed and gear controller for self‐driving vehicles.

AbstractWith the growth of urban residential density, cities have developed into metropolitan, resulting in increasingly complex individual travel behaviours. These developments pose challenges to current simulation models in activity scheduling. This paper proposed an activity scheduling and multi‐agent micro‐simulation platform (ASMMSP). By incorporating long‐term cellular data, the platform can eliminate the reliance on personal attributes in activity scheduling, which improves the simulation flexibility and accuracy. ASMMSP also focuses on transfer behaviours between different travel modes. The platform comprises three systems: agent, public transportation, and road network. At each moment, agents evaluate their current states and activity schedules, then change schedules based on the comparison results, current travel conditions, and historical travels. ASMMSP reconstructs the traffic condition within the research area by integrating the current traffic flow and activity schedules iteratively. Furthermore, ASMMSP allows for observation of real‐time traffic conditions. It also enables adjustments to public transportation, road network structure, and traffic volume, which can simulate the traffic impact from emergencies, gatherings, road maintenance, and public transportation adjustments. These functions support traffic models applied in traffic planning, development, and construction. Finally, this paper demonstrates the above capabilities through two case studies in the first ring road of Chengdu.

AbstractDue to road curvature and sensors’ limited field of view, as‐built highway curves would pose an operational challenge to the adaptive cruise control (ACC) system and its shared control. However, very few studies explored the adaptability of ACC system‐dedicated vehicles (V‐ACC) considering the vehicle‐road geometry interaction. Therefore, the objectives of this study are twofold: (i) investigating the implications of horizontal curves on V‐ACC dynamics and kinematics characteristics; and (ii) evaluating V‐ACC's adaptability from the safety, comfort, and speed consistency (S‐C‐S) aspects. To this end, a PreScan/CarSim/MATLAB/Simulink co‐simulation platform is established and it is validated by OpenACC database followed by designing many tests featuring circular curve radius (RC), desired speed (Vde), and clearance. The impact mechanism of geometric features was analysed by interpreting dynamics and kinematics characteristics along curves and critical features were further extracted by reference to S‐C‐S thresholds. The results show that: (i) either smaller RC or higher Vde causes those characteristics toward their S‐C‐S margins; (ii) neither sideslip nor rollover occurs, and speed consistency is good in most RC conditions; and (iii) drivers can follow the leading car comfortably with Vde = 40, 80–100 km/h but feel uncomfortable when Vde = 50–70 km/h and RC approaches its lower bounds.

AbstractDriving risk assessment is crucial for autonomous vehicles to guarantee driving safety and traffic efficiency. Risk field models with insufficient consideration of traffic factors are not reliable enough to provide effective support for automated driving tasks, and those highly complex models with numerous uncertain coefficients also limit the execution of automated driving tasks. Inspired by Coulomb's law, this paper proposes a new lightweight social cognitive driving risk potential field model by leveraging interaction forces between charges to explore the effects of dynamic and static traffic factors on driving risks. Through complexity analysis, the number of coefficients in the proposed model was reduced by 36%–50% compared to other models. With parametric analysis and sensitivity analysis, the model's reliability was demonstrated. A path planner was designed by integrating the proposed driving risk field model into a model predictive controller for validating the efficacy of the proposed risk potential field model. The planned path with the proposed risk field model was also compared with existing risk potential field models. Results indicate that the proposed model can effectively account for both dynamic and static traffic factors, thereby supporting the path planner to generate highly adaptable paths for complex traffic scenarios.

AbstractThe rising number of car owners has increased the frequency of drunk driving‐related traffic accidents, which is a significant danger to traffic safety. Many drawbacks of traditional drunk driving detection techniques include missed detection, interference with regular drivers, inadequate real‐time monitoring, and excessive labour costs. In this work, the intent is to increase the accuracy, real‐time performance, and coverage of drunk driving detection by proposing a method for differentiating abnormal driving conditions while intoxicated by utilizing unmanned aerial vehicle technology. The approach uses an unmanned aerial vehicle to identify the driver's facial expression to determine whether there is an evidence of drunk driving behaviour is drunk driving behaviour. It then uses these models to score vehicle trajectory anomalies, including vehicle sway, vehicle sudden speed change, and signalized intersection waiting time. According to the trial data, the system can successfully identify drunk drivers, and its accuracy has increased by 10.8% compared to the high accuracy and real‐time performance of traditional drunk driving detection methods.

AbstractFocusing on the transportation organization characteristics of the high‐speed maglev station, this article simulates the train operation process based on the train running calculation and the train operation scheme for multi‐type trains, and analyzes the station capacity based on the simulation for multi‐type trains and typical‐layout stations. This study is carried out from the following three perspectives: (i) calculating train running in the train operation process, (ii) calculating the carrying capacity for single‐type trains, and (iii) simulating the train operation scheme at the station. By analyzing the running process and working conditions of a high‐speed maglev train during various operations, a kinematic model for the train running and a simulation method for the technical operation process based on the operation calculation are established. The simulation results for train running calculation are then used to determine the time required and the capacity for the station handing single‐type trains, which is the basis of analyzing multi‐type trains operations. In the simulation for operations of multi‐type trains, three simulation strategies are designed: track selection strategy, route segment lock/unlock strategy, and operation method choosing strategy. The operational process of a high‐speed maglev station under multi‐train and multi‐technology conditions is simulated based on the above strategies, and the capacity utilization of the station is calculated. The capacity utilization characteristics of typical‐layout stations and the recommended layouts for stations under different scenarios are given for capacity analysis and optimization.

AbstractRecently, with growing interest in urban mobility patterns, the demand for collecting and analysing origin‐destination (OD) data is increasing. Due to the large scale and dimensionality of OD data, there are two issues in analysing the data: big‐data storage and major pattern extraction. To deal with two issues at the same time, this study suggests a principal control analysis‐based major demand identification method to improve the usability of microscopic OD data. Especially, this study focuses on finding principal components that preserve major patterns from OD data with small random noise so that the data can be effectively used for mobility service design. The proposed method is applied to smart card data of Seoul and Sejong and extracted major demand patterns from peak‐ and non‐peak hour data of these cities. The degree of daily regularity, reconstruction accuracy, and compression rate of the reconstructed data is analysed varying sets of principal components. The obtained results show that the major demands contain a low volume and a large volume of demand and with lower‐order principal components, major demands can be efficiently extracted by removing randomly appearing small‐volume demand. In addition, the trade‐off behaviour is observed between the degree of daily regularity and reconstruction accuracy depending on the compression rate. Based on the observations, it can be found that the loss of major demand patterns could be prevented when targeting a reconstruction accuracy of 90–95% and the proposed method can reduce the data size while preserving major mobility patterns.