Global warming demands urgent action to reduce greenhouse gas emissions, with road transport being a major contributor. Electrification of passenger cars is widely promoted as a key mitigation strategy; however, its real-world impact remains debated. This study develops a calculation framework to estimate annual fleet-level greenhouse gas emissions based on actual consumption data, vehicle type distribution, age and mileage. Using this method, short- and mid-term drawbacks of electrification are quantified alongside long-term benefits through scenario analysis aligned with the European Union 2035 target. Results indicate that electrification begins to deliver net emission reductions after 2032, achieving a 5.3% decrease by 2035 compared to an internal combustion engine fleet. These findings highlight the temporal dynamics of electrification and provide evidence-based insights for policymakers and industry stakeholders.
In the operation of railway transportation enterprises, having prior knowledge of future loading volumes and trends at freight stations is crucial for optimal deployment of empty cars and the development of daily operational plans. Long-term freight volumes at railway stations often exhibit cyclical patterns influenced by seasonal fluctuations, holidays and other factors. Additionally, changes in freight volume are significantly affected by the proportion of freight from various industries. To address these dynamics, this study proposes a hybrid prediction model for long-term loading volumes at railway freight stations. This model predicts by stage segmentation through peak-valley segmentation (PVS), variational mode decomposition (VMD) and an attention mechanism integrated into a temporal convolutional network (TCN). Using historical freight volumes from the Shuohuang railway freight station as a case study, we employed mean absolute percentage error (MAPE), mean absolute error (MAE) and root mean square error (RMSE) as evaluation metrics to assess the combined predictive performance of the PVS, VMD and TCN methods. The experimental results show that the PVS-VMD-A-TCN model significantly improves the accuracy of long-term freight volume predictions. Compared to traditional methods such as ARIMA, GRU and TCN, it exhibits superior predictive performance, offering a new approach for accurately forecasting long-term loading volumes.
With the improvement of deep learning technology, traditional obstacle avoidance approaches for blind people can no longer meet practical needs. In response to the problems of poor adaptability to multiple scenarios and low obstacle avoidance rates in current obstacle avoidance methods for blind people, this study proposes an intelligent obstacle avoidance method for blind people based on an improved You Only Look Once (YOLO) algorithm and binocular vision. This method first improves the YOLOv8 algorithm by introducing an angle loss function, SPD convolution and a more efficient local convolution structure, and proposes an I-YOLOv8-BBOD model suitable for blind people’s travel. Then, based on this model, a blind travel intelligent obstacle avoidance platform is built. The confusion matrix analysis showed that the research model performed the best and had significant advantages in reducing false positives and false negatives. Its precision, recall and F1 score were all above 0.90, indicating the best overall performance. The ablation experiment showed that after the improvement of each module, the mAP and the average accuracy at the 50% IoU threshold increased by 13.19% and 16.00%, significantly improving the detection accuracy. In the example application, the successful obstacle avoidance rate of the proposed obstacle avoidance platform exceeded 90%, and the false alarm rate was below 1.3%. The traditional platform could reach up to 5.4%, which was better than the previous platform. This indicates that the research method can accurately detect obstacles and road conditions, ensuring the safety of visually impaired people’s travel.
The long-term coexistence of connected and automated vehicles (CAVs) and human-driven vehicles (HDVs) in mixed traffic flows will continue to significantly impact merging behaviours and operational efficiency in freeway on-ramp merging areas. To improve merging efficiency at freeway on-ramps, this paper proposes a cooperative merging model for mixed traffic. First, the merging points for ramp vehicles are determined based on vehicle interaction characteristics in the merging area. Then, for the dynamic decision-making problem of merging sequences, considering the differences in cooperative merging between CAVs and HDVs, the optimal merging sequence for both single-vehicle and multi-vehicle platoon merging is determined by minimising the global deviation between the actual travel time and the expected minimum travel time. Finally, according to the predetermined merging sequence, a bidirectional speed guidance strategy based on virtual vehicle mapping is designed for both mainline and ramp vehicles to achieve cooperative optimisation of the merging sequence and vehicle trajectories, enabling the smooth merging of ramp vehicles into the mainline. Simulations verify that the proposed method performs better than FIFO and effectively improves merging efficiency under different CAV penetration rates. Particularly in low-to-medium penetration scenarios, it achieves a 25% delay optimisation rate and 22% reduction in travel time, with the optimisation effect improving to varying degrees as penetration rates increase. This establishes a foundation for traffic analysis and cooperative control in freeway on-ramp merging areas under mixed traffic environments.
This work investigates intent recognition and trajectory prediction for multiple types of traffic participants at an unsignalized intersection within connected intelligence environments based on bidirectional spatiotemporal attention network (Bi-STANet). An unsignalized intersection is used as the studied object, where the participants include Connected and Automated Vehicles (CAVs), Human Vehicles (HVs), bicyclists, and pedestrians. A novel method is proposed based on Bi-STANet. First, a multimodal spatiotemporal feature extraction model is constructed based on (2+1)-dimensional CNN ((2+1)D CNN), where grid encoding method is used to unify the spatial structure and 2D convolution is used to extract spatial features for capturing the disordered characteristics of participants. Temporal dynamics are modelled via 1D convolution along the time axis, enabling spatiotemporal decoupling. Second, a bidirectional dynamic interaction model is developed by integrating LSTM-based temporal feature extraction with (2+1)D CNN layers, where heterogeneous modality fusion is implemented through a Bidirectional Contextual Block (BiCoBlock). Finally, a model integrating dynamic interaction, intent recognition, and trajectory prediction is developed. The proposed method is validated through the inD dataset innovatively. The results show that the average accuracy of intent recognition can reach to 95.4%. Within a 3-second horizon, Average Displacement Error (ADE) and Final Displacement Error (FDE) can be reduced to 0.51 m and 0.64 m, respectively, compared with the best baseline model. In Ablation studies, intent recognition F1-score can be enhanced by 7.2%, and ADE and FDE of trajectory prediction can be enhanced by 41.4% and 39.0%, respectively.
Accurately identifying the significant factors influencing the severity of two-wheeler casualty accidents and mining association rules between multi-factor combinations and accident outcomes is crucial for implementing targeted prevention strategies. This study integrates machine learning and an improved Apriori algorithm to analyse 5,032 two-wheeler casualty accidents. Initial factors were identified through feature screening, feature fusion and K-means clustering, and then ranked by importance using the XGBoost model, resulting in ten significant factors for association rule mining. Accident severity weights were calculated based on equivalent minor injury values, and an improved Apriori algorithm, incorporating weighted support and constraints on consequent terms, was applied to extract association rules for both major and general accidents. Results show that combinations of factors such as lorries, two-wheeler riders aged 59 years or older, segregated roads, asphalt structures, electric bicycles, summer and lack of helmet use are strongly associated with major accidents. Conversely, combinations involving passenger vehicles, motorcycles, two-wheeler riders aged 30 to 43 years, non-segregated roads, cement structures and spring are associated with general accidents. Among these factors, the collision object has the greatest impact on accident severity. Minimising spatial and temporal conflicts between two-wheelers and other vehicles is essential to reducing two-wheeler casualties.
The pre-flight walk-around inspection represents a critical safety procedure in civil aviation; however, its cognitive and physiological workload mechanisms remain underexplored. This study systematically investigated pilot workload during standardised SR20 aircraft inspections by integrating heart rate variability (HRV) feature analysis, expert evaluation and task performance assessment. Forty-one flight students, stratified by training and experience levels, participated under controlled real-aircraft conditions. HRV signals were collected using wearable optical sensors, and multiple time-, frequency- and nonlinear-domain features were extracted. Feature selection was conducted via statistical testing and random forest ranking, identifying mean HR, maximum HR, standard deviation of HR, mean NNI and median NNI as the most workload-sensitive indicators. Six deep learning architectures were developed and compared under two feature-processing strategies: complete HRV features and PCA-reduced features. Results revealed a clear gradient of workload across experience groups, with experienced pilots exhibiting the lowest load and highest task performance, whereas untrained participants demonstrated the highest load. In terms of classification, the Depthwise CNN achieved the best overall performance (accuracy = 0.9372) with full HRV features, while the CNN-GRU hybrid was most effective (accuracy = 0.9186) after PCA reduction. These findings highlight the importance of aligning feature dimensionality with model architecture and provide an empirical foundation for establishing objective workload monitoring frameworks in aviation safety management.
A review of recent academic and professional literature on flight efficiency reveals that vertical and horizontal inefficiencies persist as significant challenges to the performance and sustainability of the air traffic management (ATM) system. While these inefficiencies have been widely studied, there remains a lack of comprehensive insight into their full range of impacts, as well as the underlying factors that hinder effective optimisation efforts across different operational contexts. To address this gap, this study examines and synthesises the current body of research related to vertical and horizontal flight (in)efficiency. The review synthesises findings from diverse sources, identifying recurring themes such as fuel consumption, environmental impact, monetary effects, ATM operational constraints and trajectory optimisation strategies. Additionally, the analysis underscores key differences in methodological approaches and thematic priorities between vertical and horizontal efficiency studies. The results contribute to a more holistic understanding of flight (in)efficiency and aim to inform the future development of integrated, context-sensitive models that can improve operational efficiency and support the sustainable evolution of the ATM system.
Numerous studies have demonstrated that ride-sharing optimisation for shared autonomous vehicles (SAVs) can effectively address the efficiency limitations of conventional shared mobility systems, realising the sustainable urban transportation vision of “serving more passengers with fewer vehicles”. However, current shared autonomous vehicle (SAV) ride-sharing models exhibit inefficiencies. To overcome this, we propose a real-time SAV scheduling model with dynamic detour ride-sharing, minimising total passenger travel time while maximising ride-sharing distance ratios, thus fulfilling ride-sharing demands within defined spatial ranges. To solve this model, a two-stage ride-sharing scheduling matching approach was implemented, which integrates feasible matching pair acquisition and vehicle scheduling optimisation, with robustness and opportunism evaluated via Information Gap Decision Theory. MATLAB simulations validate the model by comparing dynamic detour ride-sharing with traditional overlapping modes. Results show the dynamic detour mode achieves a 64% ride-sharing rate using 68 vehicles, outperforming conventional modes with a 22% higher ride-sharing rate and 11 fewer vehicles required. This approach enhances ride-sharing adoption, alleviates congestion, lowers travel costs and maximises passenger service efficiency with minimal fleet size.
The necessity for collaborative traffic signal control optimisation is growing in importance as intelligent transportation technology continues to advance. The study started a related application investigation to successfully raise the degree of cooperative control of traffic lights at several crossings. Based on an in-depth analysis of the traffic scenarios at multiple intersections, the study clarified the key points for optimising the collaborative control of traffic signals. Subsequently, an attention convolution model was introduced to capture the dynamic information changes of key spatial regions and data in the traffic scene through a spatial and temporal attention mechanism. The average vehicle travel time under model control was tested to be 12.4% and 23.3% shorter than the other two methods, respectively. The ablation experiments showed that the model reduced the passing time by about 9.5% and 6.7%. Moreover, the attention is mainly focused on the nodes closer to the location, and the model is more spatially interpretable. The results illustrate that the designed collaborative control of traffic signals at multiple intersections method based on the convolutional model with attention mechanism can enhance the flexibility of traffic management on the basis of improving the efficiency of collaborative control. This offers compelling evidence in favour of building a more clever and effective cooperative traffic signal control system at several crossings.
Maritime enterprises operate in a volatile and international environment, making them highly susceptible to crises and disruptions. In such conditions, organisational resilience is essential for mitigating impacts, seizing opportunities and maintaining competitive advantage. This study investigates how competitiveness strategies mediate the relationship between organisational resilience and business performance among micro, small and medium-sized enterprises (MSMEs) in the maritime sector under crisis conditions. An online survey of maritime enterprise managers was analysed using structural equation modelling (SEM) with IBM SPSS and AMOS. Findings reveal that organisational resilience has no direct effect on business performance; instead, its influence is realised through competitiveness strategies. Cost leadership, differentiation, focus and hybrid approaches allow MSMEs to leverage resilience into improved market outcomes. The results underscore the need to translate resilience into a clear strategic direction to generate measurable benefits. Differentiation and hybrid strategies, in particular, appear most effective in strengthening market position. The study also identifies corporate reputation and customer loyalty as key resilience outcomes that should be fostered in both crisis and stable periods. This research provides a rare empirical perspective on organisational resilience in the maritime sector, contributing to a limited body of literature and offering actionable guidance for managers.
As a core task of intelligent transportation systems, traffic flow prediction is characterised by high spatio-temporal complexity. Due to the limitations of existing methods in modelling complex spatio-temporal dependencies, particularly regarding medium- to long-term prediction accuracy and generalisation capabilities, this paper proposes a combination prediction model based on a multi-level dynamic GCN-Transformer framework (DGTFormer), to enhance the accuracy of short- and long-term traffic flow predictions. DGTFormer adopts a dual-stream architecture to achieve spatio-temporal decoupling modelling. The spatio-temporal dynamic graph convolutional network processes dynamic changes in the road network structure, and the temporal transformer encoder processes temporal information related to traffic flow. A spatio-temporal gated fusion mechanism is introduced to deeply couple spatial and temporal information. Experimental results on three real-world traffic datasets (PeMSD4, PeMSD7 and PeMSD8) demonstrate that DGTFormer significantly outperforms mainstream baseline models on multiple key evaluation metrics. Compared with advanced baseline methods, DGTFormer achieves performance improvements of up to 8.61% and 9.15% in RMSE and MAE, respectively. Furthermore, the coefficient of determination, R², remains stable at an excellent level above 0.9 across different prediction time steps, which fully validates that the DGTFormer model possesses superior predictive performance and generalisation capabilities.
As a significant innovation in modern transportation, autonomous driving technology offers unprecedented opportunities for improving traffic safety and travel convenience. However, under low-light conditions, insufficient illumination can cause object blurring, indistinct edges and confusion between the foreground and background, thereby compromising the reliability and safety of autonomous driving systems. Consequently, achieving robust object detection under low-light conditions has become a critical challenge in autonomous driving research. To address this issue, we propose DarkDet, a framework specifically designed for object detection in low-light environments. DarkDet consists of a Low-Light Enhancement Module (LLEM) and a Channel Fusion Module (CFM). The LLEM enhances features in dark regions of an image, while the CFM employs an attention mechanism to fuse multi-channel features, which are then fed into an object detector for final predictions. Furthermore, to expand object detection datasets tailored for low-light autonomous driving scenarios, we augment the original BDD100K dataset to construct Dark-BDD100K and collect an additional dataset, DarkCity, comprising 5,000 annotated nighttime urban road images. Finally, we evaluate the proposed DarkDet framework on Dark-BDD100K and DarkCity. Experimental results demonstrate that DarkDet achieves significant performance improvements under low-light conditions, effectively enhancing the accuracy and stability of object detection.
This study investigates the impact of individual characteristics on taxi drivers’ speed choices based on a survey of 102 valid responses. The drivers’ speed behaviour was categorised into four intervals, considering age, driving experience, personality traits and vision correction. A model based on disaggregated theory was developed to measure the impact of these factors on driving speed. Using elasticity theory, the sensitivity of these factors was analysed. The results indicated that the absolute values of the elasticity coefficients for age, vision correction and passenger presence were all below 1.000, implying that these factors have a relatively inelastic influence on the selection of speed intervals. In contrast, in the four speed intervals, the elasticity values of education are -6.194, -4.108, -5.011 and -2.972, and those of personality are 7.228, 7.602, 10.753 and 9.298; all of them have an absolute value greater than 1.000. These findings suggest that education level and personality traits significantly impact taxi drivers’ capacity to drive safely. For the benefit of passenger safety, the qualification and daily management of taxi drivers must consider their individual characteristics.
Vaccine supply chains confront several challenges in guaranteeing security of vaccines from the phase of production to the injection. Thus, it is crucial to construct a robust vaccine supply system that helps store and deliver vaccines safely and effectively. Inspired by a real-world vaccine supply case, this paper studies a robust two-echelon inventory routing problem that integrates the optimisation of vendor managed inventory (VMI), inventory policy and vehicle routing. A deterministic model is formulated to jointly optimise two-echelon vehicle routes as well as inventory at both distribution warehouse and vaccine demand nodes. Then, a robust modelling approach is introduced and converted into an equivalent linear form. An efficient solution algorithm is proposed, merging simulated annealing (SA) and adaptive large neighbourhood search (ALNS) to address large-scale cases. The computational efficiency of the proposed algorithm is demonstrated through numerical comparisons to the solutions founded by CPLEX and ALNS. The relation between input parameters and optimised results is investigated by a real-world vaccine supply system. The findings from results and sensitivity analyses suggest some interesting managerial implications.
Road damage detection is vital for effective road maintenance and ensuring traffic safety. However, existing object detection models struggle with small objects, interference from complex backgrounds and difficulty handling multi-scale object features. To tackle these challenges, this study proposes an improved road damage detection model based on YOLO11. A novel RoadRep-C3 module is introduced to improve feature extraction, while an efficient multi-scale attention (EMA) mechanism captures multi-scale damage features more effectively. Additionally, a hypergraph structure is incorporated into the neck network to enable cross-stage information fusion, improving the detection of small objects. The proposed model also utilises a slide loss function to optimise performance on challenging samples. Experimental results on the RDD2022 dataset show a 2% increase in mean average precision (mAP@0.5) over the original YOLO11, with a reduced model size. These findings demonstrate the model’s high accuracy and efficiency, offering a practical solution for detecting road damage and enhancing traffic safety.
Conventional vehicles rely on petrol and its derivatives to deliver strong performance and extended range. However, these vehicles significantly contribute to environmental pollution and the depletion of fossil fuels. Consequently, alternative-fuelled vehicle technologies have been developed, leading to a paradigm shift in the transportation sector toward electric vehicles (EVs) that reduce greenhouse gas emissions and lessen their environmental impact. For potential EV buyers, a selection problem exists due to the diverse range of alternatives. This study proposes an integrated model that selects and ranks EVs based on three main criteria: cost, driving performance and technical features. Hierarchical best worst method (HBWM), full consistency method (FUCOM), and step-wise weight assessment ratio analysis (SWARA) methods are employed to determine the importance weights for the main criteria and their sub-criteria. A sample EV set consisting of 10 potential alternatives was created for the study, and these EVs were ranked by combining the weights of the main criteria and sub-criteria obtained using the three aforementioned methods employing the technique of precise order preference (TPOP) method. Our experiments suggest that the most critical criterion among the 10 sub-criteria was price, and the least important criterion was high speed.
With the advancement of connected automated vehicles (CAVs), their increasing presence in traffic flows is expected to reshape road capacity. In urban tunnels, diverging areas are essential for traffic efficiency and safety. However, existing research has mainly focused on merging and diverging areas in expressways and arterial roads, with limited attention to urban tunnel environments. To address this gap, this study proposes a capacity estimation method for the diverging areas in urban tunnels under an intelligent connected environment. A simulation framework was developed to analyse key influencing factors under different CAV penetration rates. Then, a capacity estimation model was established using a multivariate nonlinear fitting approach and was validated through SUMO simulation experiments. The results indicate that: (1) penetration rate, diversion ratio and deceleration lane length are key parameters influencing capacity; (2) a CAV penetration rate of 0.5 marks a threshold beyond which deceleration lane length has little impact; and (3) a case study using the Nanping Tunnel in Chongqing shows that the model’s prediction error is within 5.29%. This study offers a targeted capacity estimation method for diverging areas in urban tunnels, aiding adaptive traffic control design and CAV deployment planning.
This study proposes a timetable multi-objective optimisation model for urban rail transit during the end-of-operation period, addressing dynamic passenger demand under limited train capacity. The model simultaneously minimises the unreachable rate of last-train origin-destination (OD) points and passenger flow-induced congestion at the station. Key decisions involve optimising dwell times and headways while incorporating dynamic transfer spatiotemporal constraints and train capacity limitations. To resolve the computational complexity from non-convex and non-linear terms, a multi-objective whale optimisation algorithm is employed. Case studies under large-scale events reveal that moderately extending dwell times (1.5-2 minutes) alleviates congestion, while excessive extensions prove ineffective due to line length limitations. For headways, 8-10 minutes generally optimises departure frequency, though specific Pareto-optimal solutions may flexibly adopt 4-8 minute intervals. The findings enable operators to balance accessibility and congestion mitigation based on event characteristics, enhancing operational efficiency and passenger satisfaction. The research provides actionable strategies for improving last-train transfer coordination and station flow management during special operational periods.
Road transport safety is a global concern, impacting public health, economic stability and quality of life. Understanding how economic growth and activity, represented by gross domestic product (GDP/capita), correlate with road safety outcomes is essential for crafting effective policies and interventions. This study investigates the complex relationship between gross domestic product (GDP/capita) and road safety in 27 European countries from 2005 to 2020. Using data on traffic fatalities per capita and vehicles, this research examines the influence of economic indicators on road safety outcomes, employing the original TRINCA model to frame the analysis within stages of economic development. The findings reveal that higher GDP levels are generally associated with improved road safety. However, the pace of these improvements varies significantly across European countries, reflecting diverse socio-economic contexts and regulatory frameworks. Cluster analysis classifies countries based on safety and economic performance, highlighting disparities in progress and opportunities for targeted interventions to prove the theoretical framework of the TRINCA model still exists. This study underscores the critical role of sustained economic growth and policy reform in achieving safer road systems.
