Online motorcycle taxi drivers are at high risk of traffic accidents due to human and environmental factors. Malang City, with its dense traffic and varied road conditions, presents a particularly risky setting. This study aims to analyze factors influencing traffic accidents among online motorcycle taxi drivers in Malang City. A cross-sectional study was conducted on 44 drivers using questionnaires that covered demographics, fatigue, workload, driving behavior, road conditions, and weather. Data were analyzed with Spearman’s correlation (α = 0.05). Respondents were predominantly male (88.6%) and aged ≥36 years (75%). Most reported low fatigue (61.4%) and good driving behavior (65.9%). Accident occurrence was significantly correlated with age (r = 0.370, p = 0.013), fatigue (r = 0.583, p < 0.001), workload (r = 0.402, p = 0.007), driving behavior (r = -0.442, p = 0.003), road conditions (r = 0.354, p = 0.018), and weather (r = 0.391, p = 0.009). Gender showed no association. Fatigue, a large workload, unsafe driving behavior, poor road conditions, and adverse weather conditions all increase the risk of accidents. Preventive strategies should emphasize workload regulation, safety training, and infrastructure improvement.

SemiDDM-weather: A semi-supervised learning framework for all-in-one adverse weather removal.

Image dehazing is a challenging ill-posed problem in low-level computer vision tasks, requiring the restoration of high-quality, haze-free images from complex and foggy conditions. Deep learning-based dehazing methods struggle to effectively remove non-homogeneous fog distributions due to the uneven and dense nature of fog patches, making it difficult to clear real-world fog variations. A key challenge for non-homogeneous image dehazing algorithms is efficiently capturing the spatial distribution of haze in areas with varying fog densities while restoring fine image details. To address these challenges, we propose MLCANet, a multi-level composite attention-guided network for non-homogeneous image dehazing. MLCANet mitigates the impact of uneven haze areas through two main components: the Multi-level Composite Attention Generation Network (MCAGN) and the Dehazed Image Reconstruction Network (DIRN). The MCAGN integrates channel attention (CA), spatial attention (SA), and multi-scale pixel attention (MSPA) to capture haze features at different spatial scales. The DIRN, based on a decoder-encoder architecture, combines multi-scale dilated convolutions and deformable convolutions to restore fine image details more flexibly and efficiently. Extensive qualitative and quantitative experiments, along with ablation studies, demonstrate the effectiveness and feasibility of this method for non-homogeneous image dehazing.

Innovative hive materials could reduce stress from high temperatures and adverse weather conditions for honey bee colonies, Apis mellifera L. (Hymenoptera: Apidae). This study compares traditional Langstroth wooden hives with those made from polyurethane (PU) with respect to thermoregulation and performance of four mtDNA honey bee haplotypes from June to December. The weight, internal temperature, and humidity of 10 colonies kept in PU hives and 10 colonies kept in wooden hives were continuously monitored. At the same time, brood production, Varroa infestation levels, and deformed wing virus levels were assessed once mid-experiment. No differences were observed in brood production, Varroa infestation, or viral loads between hive groups. However, PU hives maintained significantly higher average internal temperature (34.3 °C) and lower relative humidity (57.2%) compared to wooden hives (32.8 °C and 60.2%, respectively). Overall, weight loss was significantly higher in wooden hives (4.8 kg) compared to PU hives (0.8 kg), and a significant seasonal component in thermoregulation and weight was also observed. Four honey bee haplotypes were identified: C1 (A. m. ligustica), C2d and C2j (A. m. carnica), and M2-1021-7-USA (A. m. mellifera). Within each hive group, haplotype variance significantly affected colony thermoregulation and weight. Hive temperature and weight were correlated regardless of the hive material. The haplotype C1, Italian bees, a common haplotype in the USA, recorded the highest temperature and weight in wooden hives. This study demonstrated significant impacts of hive materials on honey bee colony thermoregulation and performance.

Purpose The outdoor construction performance is significantly influenced by meteorological factors. In prefabricated building projects, onsite assembly tasks, such as alignment, installation and grouting, still require substantial manual involvement, making them sensitive to adverse weather conditions. This study was aimed at examining the effect of key meteorological factors (temperature, wind, and rainfall) on the construction duration, cost and carbon emissions of prefabricated buildings. Design/methodology/approach A logical model of the standard floor construction process for prefabricated buildings was developed using Arena software. By quantifying the changes in construction duration, cost and carbon emissions under different weather conditions, the effect of the adverse meteorological factors on worker performance during the prefabricated building construction process was analysed. Findings (1) In the simulated case, the wind impact was the highest on construction performance, increasing time by 24.2%, costs by 28.6%, and carbon emissions by 29.7%, followed by high temperatures (18.0%, 21.2% and 22.2%) and rainfall (16.5%, 7.4%, 8.4%), (2) windy weather poses most pronounced risks among the three meteorological conditions, suggesting it may warrant particular attention from managers and (3) performance indicators are significantly positively correlated under different meteorological factors, which allows for the estimation of unknown indicators through established linear relationships. Originality/value These findings provide insights that may help construction companies reduce the adverse effects of meteorological factors on worker productivity. They also enable the prediction of unknown performance indicators by unlocking the linear relationship between time, cost and carbon emissions, thereby enhancing proactive decision making.

Synthetic aperture radar (SAR) imagery is widely used for target detection in complex backgrounds and adverse weather conditions. However, high-precision detection of rotated small targets remains challenging due to severe speckle noise, significant scale variations, and the need for robust rotation-aware representations. To address these issues, we propose SAR-DRBNet, a high-precision rotated small-target detection framework built upon YOLOv13. First, we introduce a Detail-Enhanced Oriented Bounding Box detection head (DEOBB), which leverages multi-branch enhanced convolutions to strengthen fine-grained feature extraction and improve oriented bounding box regression, thereby enhancing rotation sensitivity and localization accuracy for small targets. Second, we design a Ck-MultiDilated Reparameterization Block (CkDRB) that captures multi-scale contextual cues and suppresses speckle interference via multi-branch dilated convolutions and an efficient reparameterization strategy. Third, we propose a Dynamic Feature Weaving module (DynWeave) that integrates global–local dual attention with dynamic large-kernel convolutions to adaptively fuse features across scales and orientations, improving robustness in cluttered SAR scenes. Extensive experiments on three widely used SAR rotated object detection benchmarks (HRSID, RSDD-SAR, and DSSDD) demonstrate that SAR-DRBNet achieves a strong balance between detection accuracy and computational efficiency compared with state-of-the-art oriented bounding box detectors, while exhibiting superior cross-dataset generalization. These results indicate that SAR-DRBNet provides an effective and reliable solution for rotated small-target detection in SAR imagery.

This paper examines the technological aspects of mobile mini-hydroelectric power plants designed to independently provide electricityto remote mountainous areas. Key engineering solutions covering turbine types, control mechanisms, and the potential for adaptation to diverse hydrological parameters have been studied. The advantages of mobile stationsare emphasized – their small size, ease of transportation and installation,and the ability to quickly launch and integrate into localpower grids. Comparison with conventional energy sourceshas been carried out. It is shown that mini-hydroelectricpower plants can significantly reduce technical costs, increase the stability of power supply and guarantee uninterrupted operation of equipment even in adverse weather conditions. The article focuses on the economic justification and environmental safety of the use of mobile micro-hydroelectric power plants. An analysis of the costs of purchasing, installing and maintaining stations, as well as an assessment of the environmental impact, is presented. It is noted that with proper planning and use, micro-hydroelectricpower plants have little impact on water resources and ecosystemsof mountain rivers.

Abstract. The study presents a robust, automated camera gauge for long-term river water level monitoring operating in near real-time. The system employs artificial intelligence (AI) for the image-based segmentation of water bodies and the identification of ground control points (GCPs), combined with photogrammetric techniques, to determine water levels from surveillance camera data acquired every 15 min. The method was tested at four locations over a period of more than 2.5 years. During this period almost 218 000 images were processed. The results demonstrate a high performance, with mean absolute errors ranging from 0.96 to 2.66 cm in comparison to official gauge references. The camera gauge demonstrates resilience to adverse weather and lighting conditions, achieving an image utilisation rate of above 95 % throughout the entire period. The integration of infrared illumination enabled 24/7 monitoring capabilities. Key factors influencing absolute error were identified as camera calibration, GCP stability, and vegetation changes. The low-cost, non-invasive approach advances hydrological monitoring capabilities, particularly for flood detection and mitigation in ungauged or remote areas, enhancing image-based techniques for robust, long-term environmental monitoring with frequent, near real-time updates.

Maritime target detection under complex adverse weather conditions (e.g., fog, rain, and low light) is crucial for Unmanned Surface Vehicle (USV) navigation. However, achieving high detection accuracy and efficiency remains challenging due to coupled environmental interference and limited computing resources. In this paper, we propose W-MTD, a task-specific distillation framework designed for weather-robust and lightweight maritime target detection based on knowledge distillation. Building upon the Fine-grained Distribution Refinement (D-FINE) detection model, this method constructs a dual-path knowledge distillation framework tailored for maritime scenes. Through the synergistic optimization of feature similarity constraints and decoupled distillation, it facilitates multi-level knowledge transfer from a teacher model to a lightweight student model, mitigating feature degradation caused by model compression. A multi-scenario augmentation strategy is designed to balance convergence across different weather conditions. Experiments show that W-MTD’s student model improves detection accuracy by 7.0–13.9% under three adverse weather conditionscompared to the baseline teacher model trained solely on clear weather data while maintaining comparable clear-weather performance. With only 4 M parameters and 7 GFLOPs, the student model demonstrates favorable performance and efficiency compared to other real-time detectors, indicating its potential suitability for USV deployment.

Semantic segmentation in adverse weather conditions presents significant challenges due to insufficient image brightness, excessive noise, and blurred object boundaries, which hinder the performance of traditional visual recognition methods. Domain generalization (DG) for semantic segmentation aims to leverage data from normal illumination domains to ensure robust model performance in unseen adverse weather domains-a critical requirement for autonomous driving robots. Recent advancements in parameter-efficient fine-tuning via frozen vision foundation models offer new avenues for DGs. However, conventional domain-generalized semantic segmentation methods often struggle with severe weather conditions, particularly in capturing object details and global structures. To overcome these limitations, we introduce RFGLNet, a domain-generalized semantic segmentation model designed for adverse weather scenarios. RFGLNet enhances segmentation accuracy by incorporating an SVD-Initialized Low-Rank Module, a Fourier-Enhanced Channel Attention Module, and a Grouped Modeling Spatial Attention Module. By leveraging frequency-domain information through Fourier transforms, RFGLNet improves global structural perception, facilitating a holistic understanding of complex scenarios. Additionally, the decompositional modeling spatial attention mechanism reduces cross-channel interference, enhancing local detail extraction. Using singular value decomposition for parameter fine-tuning ensures precise and rapid alignment with pretrained feature distributions. Our experiments show that RFGLNet achieves a mean intersection over union of 78.3% on the ACDC adverse weather test dataset, with only 4.32M trainable parameters.

Tactical platooning enables autonomous vehicles to operate in tightly coordinated formations, which is especially critical in military scenarios where they must execute evasive maneuvers and synchronized movements to avoid enemy threats in complex battlefield environments. Because this operation involves real-time data sharing with minimal latency to support the continuous exchange of large volumes of sensor data and control signals, stable and high-capacity links between vehicles are essential. To address this need, we propose a beam tracking technique to facilitate efficient inter-vehicle directional communication, particularly in highly dynamic and mobility-intensive situations. LiDAR (Light Detection and Ranging) is employed to accurately track the position data of the leading vehicle at a high frequency. To ensure robustness and reliability under adverse weather conditions where LiDAR measurements may be degraded or unavailable, RADAR (Radio Detection and Ranging) is incorporated into the UKF (Unscented Kalman Filter) as an alternative sensing modality. Additionally, a beam selection method is integrated to enhance the performance of the proposed technique. To verify its effectiveness, we developed a vehicle mobility and wireless channel model and random-sample simulation were conducted.

ABSTRACT Unmanned Aerial Vehicles (UAVs) play a crucial role in search and rescue (SAR) operations, surveillance, amongst others, but their deployment in adverse weather conditions, mainly high‐winds, remains a challenge. The proposed hybrid control system serves as a step toward more robust and intelligent UAV solutions capable of operating in extreme weather conditions. The advent of artificial intelligence has encouraged the move towards the development of adaptive systems, which rely on static datasets and limited generalization in complex environments, encouraging the shift towards dynamic, real‐time learning methods. This paper presents a hybrid control strategy for enhancing UAV stability in strong wind conditions by integrating Linear Quadratic Regulator (LQR) control, Particle Swarm Optimization (PSO), and Deep Deterministic Policy Gradient (DDPG) reinforcement learning. The proposed approach is implemented on the Field‐based Autonomous LiDAR Control for Obstacle Navigation (FALCON) Digital Twin, developed at Ulster University, enabling simulation and testing of wind disturbances as faults. The performance and complexity analysis of the strategies indicates that the proposed DDPG to LQR‐PSO controller (HyLPD) offers superior stability, effectively mitigating wind‐induced deviations while maintaining manageable computational complexity. Results show that while LQR‐PSO achieves a 40% performance improvement with a low computational cost, the HyLPD method further enhances system robustness, reducing overshoot and improving settling times under turbulent conditions without extensively labeled data. However, the increased computational demand of hybrid controllers suggests a trade‐off between adaptability and real‐time feasibility for UAV deployment. This study highlights the importance of hybrid control frameworks in UAV applications, particularly for SAR, where resilience to environmental disturbances is critical.

Image deraining is a crucial preprocessing task for enhancing the robustness of high-level vision systems under adverse weather conditions. However, most of the existing methods are limited to a single RGB color space, and it is difficult to effectively separate high-frequency rain streaks from low-frequency backgrounds, resulting in color distortion and detail loss in the restored image. Therefore, a rain removal network that combines dual-color space and frequency domain priors is proposed. Specifically, the devised network employs a dual-branch Transformer architecture to extract color and structural features from the RGB and YCbCr color spaces, respectively. Meanwhile, a Hybrid Attention Feedforward Block (HAFB) is constructed. HAFB achieves feature enhancement and regional focus through a progressive perception selection mechanism and a multi-scale feature extraction architecture, thereby effectively separating rain streaks from the background. Furthermore, a Wavelet-Gated Cross-Attention module is designed, including a Wavelet-Enhanced Attention Block (WEAB) and a Dual Cross-Attention module (DCA). This design enhances the complementary fusion of structural information and color features through frequency-domain guidance and bidirectional semantic interaction. Finally, experimental results on multiple datasets (i.e., Rain100L, Rain100H, Rain800, Rain12, and SPA-Data) demonstrate that the proposed method outperforms other approaches.

This paper investigates the feasibility of integrating a broadband matching device (BMD) with a high-frequency switch into the antenna system of the MIC RL-400M radio relay station, part of the "ROSA" radar complex. The aim is to compensate for antenna impedance changes caused by adverse weather conditions such as snowfall, icing, and wet snow, which reduce power transmission efficiency. Various types of high-frequency switches, including relays, PIN diodes, and transistors, were analyzed. A transistor-based switch (HMC349AMS8G) was selected due to its low insertion loss, high reliability, and wide operating frequency range. The BMD structure was synthesized to minimize impedance variation, and its performance was evaluated through simulation in AWR Microwave Office and experimental measurements in the 394–450 MHz range. Results showed an average power loss reduction of 0.7 dB and a 6.9% increase in radio link range compared to operation without a switch. The proposed solution enhances the stability and efficiency of radar and radio relay systems, ensuring reliable operation in challenging environmental conditions.

LiDAR point clouds segmentation in adverse weather conditions

The main objectives of the study focused on identifying the physical mechanisms of acoustic noise generation by high-voltage power lines under conditions of high humidity and quantitatively assessing associated energy losses. To achieve these objectives, the following tasks were accomplished: a physical-mathematical model was developed considering two complementary mechanisms - the motion of polarized water droplets in the non-uniform electric field of the wire and their subsequent destruction upon contact with the conductor; calculations were performed of the electric field strength near the wire, induced dipole moment of droplets, and the acting force; an assessment was made of droplet impact velocity on the wire and conditions for their micro-explosive destruction; and a methodology was developed for calculating additional leakage currents and power losses. The most important results are the theoretical substantiation of a new combined physical mechanism for noise generation, based on droplet polarization, acceleration, and micro-explosive destruction, and the development of a methodology for quantitative assessment of additional energy losses. The significance of the obtained results lies in proposing a comprehensive physical explanation of the acoustic phenomenon that establishes a connection between power line noise characteristics and electrophysical processes in the surface area under conditions of high humidity, as well as identifying a new mechanism of energy losses that is essential for optimizing operational regimes of high-voltage power transmission lines. The scientific novelty of the work is the proposal of this new mechanism and the established analytical relationships between key parameters. The practical significance lies in the developed methodology for assessing additional losses, which is important for improving the accuracy of loss forecasting and optimizing line operation in adverse weather conditions.

Mountain running represents a complex form of physical activity that combines athletic performance, mental endurance, and the appreciation of natural landscapes. Objectives: This paper analyzes the organization and outcomes of the “Secret Paradise” mountain cross-country race, held on October 4th, 2025, in the Bârgău Mountains, around Zimbroaia Peak (1345 m). The 16 km course, featuring a total elevation gain and loss of 1000 m, brought together 39 participants (27 men, 12 women) aged between 14 and 73. Methods: The analysis is based on direct observation, organizational data, and participants’ feedback. Results: Results highlight the potential of such events to foster community cohesion, promote active tourism, and encourage outdoor physical activity. Conclusions: Despite adverse weather conditions, the combination of a challenging course, effective organization, and scenic landscapes ensured the event’s success. Article history: Received 2025 October 13; Revised 2025 December 20; Accepted 2026 January 05;Available online 2026 January 30; Available print 2026 January 30

The article provides a comprehensive overview of modern technologies for detecting airborne objects. It emphasizes the use of machine learning (ML) methods to improve the accuracy and reliability of surveillance systems. The study is relevant due to the growing number of airborne objects that threaten civil and military infrastructure. Traditional detection methods also have limitations in adverse weather conditions or when radio interference is present. Intelligent technologies, capable of recognizing object behavior patterns and functioning under uncertainty, are especially important in this context. The primary approaches to detecting aerial objects are analyzed, including radio frequency monitoring, radar analysis, electro-optical and thermal imaging surveillance, and acoustic monitoring. Their advantages, limitations, and specific applications are identified. Special attention is paid to systems that use ML algorithms for automated object recognition and classification. Thanks to its ability to independently detect patterns in data without human intervention, machine learning ensures high accuracy and reduces the influence of the human factor. The paper summarizes the results of recent studies that demonstrate the accuracy of detection in test environments, the ability of systems to adapt to new scenarios, and the improvement in reliability through multisensor integration. It is shown that combining data from different sensors – radar, EO/IR, acoustic, and radio frequency – in conjunction with ML algorithms creates the basis for building adaptive airspace monitoring systems. The conclusions emphasize that the effectiveness of combating air threats depends on the ability of systems to integrate heterogeneous data and use intelligent analysis algorithms, which paves the way for the creation of accurate, flexible, and reliable means of detecting new types of airborne objects.

Vehicle detection is crucial for intelligent decision support in transportation systems. However, real-time detection of vehicles is challenging due to geometric variations of vehicles and complex environmental factors such as light conditions and weather. To address these issues, the paper introduces the You-Only-Look-Once with Deformable Convolution and Cross-channel Coordinate Attention (YOLO-DC) framework that improves the performance and reliability of vehicle detection. First, YOLO-DC incorporates Cross-channel Coordinate Attention, which combines channel attention and coordinate attention, to more accurately cover target sampling positions and enhance feature extraction from vehicles of various shapes. Second, to better handle vehicles of different sizes, we employ Multi-scale Grouped Convolution to enable multi-scale awareness and streamline parameter sharing. Additionally, we incorporate channel prior convolutional attention so that the model can concentrate on areas of vehicles that are critical for detection. We also optimize feature fusion by leveraging a highly efficient fusion of C2f (CSP Bottleneck with 2 Convolutions) and FasterNet to reduce the model size. Experimental results demonstrate that YOLO-DC performs better than the state-of-the-art YOLOv8n method in detecting small, medium, and large-sized vehicles, and in detecting vehicles in adverse weather conditions. In addition to its superior performance, YOLO-DC also features fast detection speed, making it appropriate for real-time detection on devices with limited computational power.

Background: The Port of Durban plays a pivotal role in South Africa’s maritime trade. This article addresses a critical challenge faced by the Durban dry-bulk and break-bulk terminal, which is the excessive waiting time experienced by vessels at anchorage. Lengthy anchorage waiting time results in economic losses for shipping companies and affects the overall efficiency of the port. Objectives: The study aimed to identify the factors that affect the anchorage of dry-bulk and break-bulk vessels while waiting for the berth (WFB) in order to recommend actions that should be taken to reduce vessel anchorage waiting time while WFB at the Port of Durban’s bulk terminal. Method: Qualitative data were obtained through semi-structured interviews, which lends itself particularly well to thematic analysis. Results: The research findings reveal that multiple factors contribute to the extended waiting times experienced by dry-bulk and break-bulk vessels at the Port of Durban. These factors include inefficient cargo handling processes, inadequate infrastructure, congestion, adverse weather conditions, and port capacity constraints. In addition, vessels arriving at the same time while the berth is still occupied further intensify delays, resulting in vessels WFB at anchorage. Conclusion: The study identified and prioritised these factors, which enabled the development of targeted solutions to mitigate the waiting time issue. The proposed solutions encompass both short-term and long-term measures. Contribution: This research emphasises the importance of supply chain collaboration among key port stakeholders.