For the problems of weak feature representation, significant scale variation, and background interference in small target features of unmanned aerial vehicle (UAV) aerial images, existing detection methods struggle to achieve both lightweight deployment and detection accuracy. Therefore, this paper proposes an extremely lightweight and accurate small target detection architecture named GS-YOLO. Through modular innovation, it achieves extreme lightness and improved detection performance. The core innovations include: 1) Design of a lightweight small target perception attention fusion module C2FGhostLight, using proportionally optimized GhostConv to replace traditional convolution, combined with a dual-path lightweight attention mechanism, which significantly reduces parameters while dynamically suppressing background interference; 2) Proposal of a lightweight channel attention module for small target perception (SOLCA), through a “channel focusing-local enhancement” dual-branch compact structure and adaptive weighted fusion, to strengthen weak feature representation. Experimental results show that on the VisDrone public dataset, GS-YOLO improves mAP50 by 0.9% compared to YOLOv8n, with a model parameter size of only 0.84M. It maintains lightweight characteristics and provides a solution for engineering applications in UAV aerial photography scenarios.

Three modes, three profiles: Characterizing fatal crashes on e-scooters, e-bikes, and conventional bicycles in Sweden.

Aerodynamic characteristics of a road vehicle passing by bridge tower areas under interfered crosswind

Study on the dynamic response and noise characteristics of metro vehicles with resilient wheels

The process of designing, manufacturing and operating automotive products in modern conditions is unthinkable without the introduction of computer technology at all stages of the life cycle. One of the key tasks in this case is to confirm the compliance of the product being developed with the requirements imposed on it using validated computational models at the earliest stages of design. The article uses several examples to show the experience of using computational studies of the aerodynamic characteristics of vehicles and their components. The obtained calculation results are compared with the available experimental data. Recommendations have been developed to improve the accuracy of calculations for estimating fuel consumption.

A detailed dataset of motor insurance policies with coverage-specific financial information

This study aims to examine factors associated with self-reported crash involvement among drivers in Pakistan using interpretable machine learning (ML) techniques and established driver-behavior instruments. The data used in this study were collected through an Internet-based survey in Pakistan utilizing a self-report questionnaire (cross-sectional design; N:623 drivers). The questionnaire consisted of items related to dangerous (aggressive, risky, and negative emotions), aberrant (errors and violations), and positive driving behaviors, along with demographics and items related to self-reported crashes (Yes/No) over the past three years. Interpretable ML techniques, including Logistic Regression, Categorical Boosting (CatBoost), and eXtreme Gradient Boosting (XGBoost) were employed. Furthermore, to improve the interpretation of ML techniques for self-reported crash risk factors, the Shapley Additive explanation (SHAP) technique was used. The results revealed that aggressive driving and risky driving behaviors, errors, and violations committed while driving, were associated with increased traffic crashes. Prior motorcycle riding experience variable also had a significant association with the occurrence of traffic crashes. Moreover, XGBoost outperformed the CatBoost and Logistic Regression models in terms of predictive performance. XGBoost achieved the highest test accuracy (0.86), compared to CatBoost (0.80) and Logistic Regression (0.73). The study underscores the importance of integrating interpretable ML in traffic safety research. The findings suggest that behavior-oriented road safety interventions such as aggressive and risky driving enforcement, driver education, and risk perception should be prioritized and enhanced. Prior motorcycle-riding experience is important, transition-oriented training should be designed for riders to adapt vehicle characteristics and road usage patterns.

This paper presents HydroAir, a novel air-propelled Unmanned Surface Vehicle (USV) specifically designed for operation in shallow waters and obstacle-rich aquatic environments such as lakes, reservoirs, and large dams. Unlike conventional aquatic robots, HydroAir employs an aerial propulsion system that enables it to overcome partially submerged obstacles, vegetation, and extremely shallow regions where traditional propeller-based platforms fail. The vehicle features a system with a very reliable internal architecture, providing high maneuverability and robustness in both manual and autonomous navigation modes. The primary objective of HydroAir is to serve as a mobile sensing platform for three-dimensional reconstruction of aquatic environments, particularly the underwater terrain. The onboard sensing suite enables bathymetric data acquisition, while a dedicated monitoring and control software integrates these data with aerial reconstructions obtained from Unmanned Aerial Vehicles (UAVs), allowing for the fusion of above-water and underwater spatial information into a unified 3D model. Experimental validations were conducted in large-scale, real-world environments, including tests in a hydroelectric dam operated by Santo Antônio Energia on the Madeira River in Brazil, demonstrating the platform’s operational feasibility, stability, and reconstruction capabilities. The results indicate that HydroAir is a promising solution for environmental monitoring, inspection, and mapping in challenging aquatic environments where conventional autonomous surface vehicles are limited.

This study explores sustainable energy management approaches for a smart distribution network that combines multiple infrastructures, such as electric vehicle charging stations, hydrogen refueling facilities for fuel cell vehicles, and renewable energy systems integrated with hydrogen storage. These components are managed in a coordinated manner to satisfy both operational requirements and security criteria defined by the distribution system operator. A key feature of the hydrogen storage unit is its dual functionality, as it not only stores electrical energy but also supplies hydrogen to end users. The primary objective is to reduce overall energy losses within the distribution system. To accomplish this, the research considers several important factors, including AC power flow modeling, grid voltage operational and security constraints, system flexibility, environmental restrictions, operational characteristics of electric vehicles charging and hydrogen stations, and performance models of renewable energy systems coupled with hydrogen storage. Furthermore, the proposed framework accounts for uncertainties related to load demand, renewable generation, and variations in the number of electric vehicles by applying a scenario-based stochastic optimization technique. The findings demonstrate significant enhancements in both system performance and security. In particular, the proposed method decreases voltage deviations, power losses, and peak load capacity by approximately 24.4%, 32.8%, and 38.3%, respectively, compared to conventional load flow analyses. Moreover, voltage security within the network is improved by nearly 10.2%, confirming the efficiency of the proposed integrated energy management strategy.

Objectives The frequent acceleration, deceleration and lane change of vehicles in the freeway ramp area lead to significant differences in the accident mechanism of the diverging, merging, and weaving zones. This study aims to analyze the heterogeneity and transferability of factors affecting the severity of accident injuries in different ramp sub-areas, and to make a comparative analysis of the significant influencing factors in different sub-areas. Methods Based on the accident data of freeway ramps in the state of Florida, the United States, this study divides the severity of accident injuries into three categories. Random parameter logit models with heterogeneity in both means and variances are developed for the overall ramp area as well as for diverging, merging, and weaving subareas. Choose 22 explanatory variables from the four aspects of driver, vehicle, roadway, and environmental characteristics. The log- likelihood ratio tests are used to evaluate the transferability of different ramp areas model parameters, and the heterogeneity of influencing factors is analyzed through the distribution characteristics of random parameters. At the same time, the direction and degree of influence of significant variables on the severity of accident injury are quantified by using the average elastic coefficient. Results The results show that there is no transferability of parameters between the overall model and subarea models or among the three ramp subareas, which verifies the necessity of zoning modeling analysis. Different ramp subareas show significantly different means and variances characteristics. “Hit and run: No,” “Pre-crash speed: [20, 40),” and “Driver gender: Male” emerge as random parameters in diverging, merging, and weaving areas. Comparative analyses further reveal that some variables exert opposite effects on crash injury severity across different ramp zones, whereas others are significant only within a single subarea. Only a limited number of factors—such as driver condition, driver gender, low pre-crash speed, and vehicle type—consistently influence crash injury severity across all ramp subareas. Conclusions The factors affecting the severity of accident injuries in the freeway ramp area have significant spatial heterogeneity. The research results systematically analyzed the significant influencing factors of each subarea, and found that driver behavior, pre-collision speed, road speed limit, lighting conditions all have different effects in different subareas. Regionally differentiated improvement measures and management policies should be formulated for different subareas.

To address the challenges of insufficient feature fusion and incomplete multi-scale information capture in complex traffic scenarios, we propose a vehicle type recognition network based on feature comparison and the Mixture of Experts (MoE) model. Specifically, the MobileNetV4 backbone is introduced to enhance deep feature extraction for vehicle targets. Meanwhile, we design a Multi-scale Interleaving Fusion Module (MSIFM), which progressively transmits feature channels via an interleaving structure to capture multi-scale features while enhancing vehicle feature representation. Moreover, we devise a Feature Compare Enhancement Module (FCEM) to efficiently fuse feature maps with different semantic information. By performing feature comparison, it strengthens strongly correlated features while suppressing weakly correlated ones. Finally, we design a Mixture of Experts Feature Enhancement Module (MOEFEM) to aggregate multi-scale feature maps and adaptively capture detailed vehicle features through multiple expert units. Experimental results demonstrate that our method achieves mAP improvements of 2.2% and 2.4% over YOLOv11 on UA-DETRAC and BDD100K, respectively. The proposed method not only improves detection accuracy significantly but also maintains real-time efficiency, providing a practical solution for high-precision vehicle type recognition. It offers valuable technical support for intelligent transportation systems, smart city management, and autonomous driving safety.

To address the poor mechanical adaptability of conventional equipment to 40 cm narrow-row sesame cultivation and the high weeding resistance and energy consumption of traditional weeding tools, this study developed an integrated bionic weeding and plant protection vehicle. The vehicle features a modular structure capable of three-row weeding and four-row plant protection, coupled with an extended-range hybrid powertrain. Its parallel linkage design enables terrain adaptation, ensuring consistent weeding depth of 3–6 cm and stable spraying height. Combined with an adjustable spraying width and a “detection–feedback–adjustment” mechanism to prevent plant collisions, the vehicle is fully compatible with the agronomic requirements of narrow-row cultivation. Inspired by mole cricket forelegs, the vehicle’s bionic weeding wheel blade model incorporates quantified biological features: quadratically fitted claw toe contours (R2 > 0.97), a toe base height-to-width ratio of 1:2, and a toe groove radius-to-toe height ratio of 1:1. This design achieves a reliable biological-to-engineering translation. EDEM-based Discrete Element Method (DEM) simulations confirm that the bionic wheel outperforms conventional designs: the average torque is 17.4% lower (7.75 vs. 9.38 N·m), the soil disturbance rate is 8.2 percentage points higher (95.2% vs. 87.0%), and soil particle motion is more ordered (average velocity: 0.52 vs. 0.58 m/s), effectively reducing energy waste and improving weeding efficiency.

Analysis of energy consumption and emissions characteristics of plug-in hybrid electric vehicle (PHEV) under various real-world driving conditions.

Conceptual design of a surface-underwater unmanned vehicle and fluid characteristics analysis during surface navigation

A self-learning refined model and tracking for near space hypersonic vehicle by space-based radar

Autonomous driving is an emerging technology that has developed rapidly over the last decade, with decision-making remaining a critical challenge, particularly due to its significant role in traffic congestion. In this paper, we propose a novel perception-and-decision framework, called <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">HEAD</i>, which consists of an en<underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">H</u>anced p<underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">E</u>rception module and a m<underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">A</u>neuver <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">D</u>ecision module to address this challenge. In the enhanced perception module, a graph-based state prediction model with a strategy of phantom vehicle construction is proposed to address incomplete vehicle features and predict future states in parallel. Then in the maneuver decision module, a deep reinforcement learning-based model is designed to learn a driving policy based on a parameterized action Markov decision process. A hybrid reward function takes into account aspects of safety, efficiency, comfort, and impact to guide the autonomous vehicle to make optimal maneuver decisions. To make our framework applicable to more scenarios, we further propose an improved <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">HEAD</i> (<italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">HEAD</i>++) framework that makes the autonomous vehicle adapt to various road structures, such as lane merging and diverging scenarios. Besides, we develop a style tuning module in <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">HEAD</i>++, which supports personalized driving style tuning. To mitigate high training costs, an efficient style tuning method with approximate gradient descent is proposed to reduce the number of training iterations. Extensive experiments demonstrate the effectiveness of our framework. Compared to state-of-the-art methods, <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">HEAD</i>++ reduces overall traffic disturbance by 23.3%-40.9%, lowers collision risk by 4.5%-17.8%, and improves passenger comfort by 13.1%-30.5%, while maintaining high traffic efficiency.

Blended-wing–body (BWB) has emerged as a potential concept to replace the traditional tube and wing (TAW) configuration. As with traditional flying wings, the BWB is prone to the wing-rock phenomenon but with a different triggering mechanism, which causes significant flight stability and control challenges. This paper aims to investigate the wing-rock characteristics of a BWB unmanned combat aerial vehicle and further evaluate the efficacy of active flow control techniques for its suppression. A validated computational framework has been developed based on rigid-body single-degree-of-freedom (single-DOF) dynamic mesh motion and forced roll sliding mesh motion employing the unsteady Reynolds-averaged Navier–Stokes equations. Free-to-roll simulations have predicted the onset angle of attack and various wing-rock characteristics. Jet blowing was influential in suppressing wing-rock amplitude and mean roll angles within a specific range of angles of attack, after which its momentum coefficient has to be increased. Liutex-based flow analysis revealed complex tip-separated flow interactions and the coalescence of multiple vortex systems as the primary causes of wing-rock initiation. The developed framework can be extended to multi-DOF analyses, flow-adaptive blowing, or to investigate other dynamic instabilities.

Purpose Humanitarian organizations (HOs) continue to lag in adopting the benefits of vehicle-sharing during disaster relief operations. Therefore, the primary objective of this study's investigation is to scrutinize and rank the important challenges that hinder vehicle-sharing initiatives among HOs. Design/methodology/approach This study employs a multi-method approach, incorporating both interviews and surveys. Initially, interviews are conducted with officials from HOs to explore the challenges associated with vehicle-sharing. Next, surveys are administered to officials of HOs to rank these identified challenges. This study is framed through the perspective or lens of innovative resistance theory. Findings The absence of government support and incentives emerges as the most critical challenge. This is followed by the unwillingness of top management and severe inter-organizational competition, which are the next most significant challenges. Subsequent challenges, in order of importance, include lack of trust, corruption and unethical behavior, lack of internal coordination, absence of a neutral third party, unawareness of potential benefits, varying characteristics of vehicles and compliance standards. Research limitations/implications By ranking these important challenges, the study offers a framework for prioritizing efforts and resources for interorganizational collaboration. The findings emphasize the need for policy interventions and strategic initiatives to address challenges to vehicle-sharing among HOs. Practical implications Key obstacles like the lack of government support, management unwillingness and inter-organizational competition highlight the importance of fostering collaboration. Originality/value The findings provide the first insights into vehicle-sharing challenges among HOs.

Under the trend of AI-defined vehicles, multi-agent collaboration has become the core feature for intelligent vehicles to deliver superior user experience (UX). Traditional linear and independent evaluation methods can no longer adapt to the new technical characteristics and logic. Taking the agents of four functional domains—intelligent driving, intelligent cockpit, intelligent vehicle control, and intelligent connectivity—and their cross-domain collaborative relationships as research objects, this study constructs a UX evaluation index system consisting of five primary indicators and 14 secondary indicators. Innovatively, the analytic network process is adopted for indicator weight allocation, which effectively characterizes the interdependencies among indicators caused by multi-agent collaboration. Meanwhile, the coupling coordination theory is introduced to construct a comprehensive UX index, enabling quantitative evaluation of the balanced development level across the five dimensions. The results show that in intelligent vehicle UX, excellence in a single dimension does not equal excellent overall UX. Only through the collaborative upgrading of multiple agents and balanced development of the five dimensions can the comprehensive UX be maximized. This study further reveals the UX mechanism of multi-agent collaboration in intelligent vehicles and determines the optimal collaborative evolution path based on the dynamic programming algorithm, providing theoretical support and practical guidance for automakers in rational product development planning.

The article contains the analysis of interrelations in the system "car—road—environment", which are realized through the interaction of working elements of attachments and machine wheels with objects of labor and road surface in specific operating conditions, manifested through the characteristics of the external environment, and form a complex multilevel system with a hierarchical principle of construction and functioning. To assess the mutual influence of changes in the characteristics of cars and the external environment, the method of non-expert evaluation is proposed, which has already found its wide application in assessing the technical level of machines.