Vehicle Classification Research Articles

Unmanned Aerial Vehicle Classification Using Neural Neworks and Radar Digital Twins: UAV Classification Using Neural Networks and Radar Digital Twins

Unmanned Aerial Vehicle Classification Using Neural Neworks and Radar Digital Twins: UAV Classification Using Neural Networks and Radar Digital Twins

Enhancing Situational Awareness with VAS-Compass Net for the Recognition of Directional Vehicle Alert Sounds

People with hearing impairments often face increased risks related to traffic accidents due to their reduced ability to perceive surrounding sounds. Given the cost and usage limitations of traditional hearing aids and cochlear implants, this study aims to develop a sound alert assistance system (SAAS) to enhance situational awareness and improve travel safety for people with hearing impairments. We proposed the VAS-Compass Net (Vehicle Alert Sound–Compass Net), which integrates three lightweight convolutional neural networks: EfficientNet-lite0, MobileNetV3-Small, and GhostNet. Through employing a fuzzy ranking ensemble technique, our proposed model can identify different categories of vehicle alert sounds and directions of sound sources on an edge computing device. The experimental dataset consisted of images derived from the sounds of approaching police cars, ambulances, fire trucks, and car horns from various directions. The audio signals were converted into spectrogram images and Mel-frequency cepstral coefficient images, and they were fused into a complete image using image stitching techniques. We successfully deployed our proposed model on a Raspberry Pi 5 microcomputer, paired with a customized smartwatch to realize an SAAS. Our experimental results demonstrated that VAS-Compass Net achieved an accuracy of 84.38% based on server-based computing and an accuracy of 83.01% based on edge computing. Our proposed SAAS has the potential to significantly enhance the situational awareness, alertness, and safety of people with hearing impairments on the road.

Evaluation of potential electric vehicles load-induced damage on flexible pavements

The adoption of Electric Vehicles (EVs), driven by policies like the Zero Emission Vehicle (ZEV) mandate, is rapidly transforming transportation infrastructure. In 2021, over 1.25-million light-duty EVs were registered in the United States, signaling a shift that extends to heavy vehicle categories. This study evaluates the implications of this transition on road networks, focusing on the increased gross weight of EVs compared to traditional Internal Combustion Engine (ICE) vehicles and its impact on flexible pavement structures. The study investigates the deterioration patterns and potential damage to road infrastructures resulting from integrating EVs into the traffic mix, utilizing both the AASHTO 1993 and Mechanistic-Empirical (ME) pavement evaluation methodologies. The analyses reveal a significant acceleration in road degradation, necessitating urgent consideration of EV-induced stresses by transportation authorities. Recommendations for future research and practical strategies for mitigating these impacts are provided to guide policymakers and engineers in adapting to this evolving vehicular landscape.

Analysis of Energy Efficiency Parameters of a Hybrid Vehicle Powered by Fuel with a Liquid Catalyst

A notable trend in the modern automotive market is the increased interest in hybrid cars. Hybrid cars combine a standard internal combustion engine with an electric motor solution. Research into increasing the energy efficiency of a conventional unit while meeting increasingly stringent exhaust emission standards is becoming a key postulate in this matter. This article discusses an analysis of modifying the fuel used by hybrid vehicles using the example of a selected drive unit equipped with a spark-ignition engine. This effect was tested after the Eco Fuel Shot liquid catalyst was added to the fuel. The research process was carried out in two stages, as follows: in road conditions using the Dynomet road dynamometer; and on the V-tech VT4/B2 chassis dynamometer. Tests were carried out to replicate road tests with a catalytic additive in the fuel. A mathematical model was created and the following energy efficiency parameters of the hybrid vehicle were calculated: the torque of the internal combustion engine, electric motor, and generator; the rotational speeds of the internal combustion engine, electric motor, and generator; the power of the internal combustion engine, electric motor, and generator; the equivalent fuel consumption of the electric motor and generator; the fuel consumption of the internal combustion engine, electric motor, and generator; and the mileage fuel consumption of the internal combustion engine, electric motor, and generator. The results of the tests made it possible to identify the benefits of using the tested liquid catalyst on the operation of the drive system of the analyzed hybrid vehicle. This research will be of benefit to both the demand side in the form of users of this category of vehicles, and the supply side represented by the manufacturers of power units.

Enhanced Vehicle Logo Detection Method Based on Self-Attention Mechanism for Electric Vehicle Application

Vehicle logo detection plays a crucial role in various computer vision applications, such as vehicle classification and detection. In this research, we propose an improved vehicle logo detection method leveraging the self-attention mechanism. Our feature-sampling structure integrates multiple attention mechanisms and bidirectional feature aggregation to enhance the discriminative power of the detection model. Specifically, we introduce the multi-head attention for multi-scale feature fusion module to capture multi-scale contextual information effectively. Moreover, we incorporate the bidirectional aggregation mechanism to facilitate information exchange between different layers of the detection network. Experimental results on a benchmark dataset (VLD-45 dataset) demonstrate that our proposed method outperforms baseline models in terms of both detection accuracy and efficiency. Our experimental evaluation using the VLD-45 dataset achieves a state-of-the-art result of 90.3% mAP. Our method has also improved AP by 10% for difficult samples, such as HAVAL and LAND ROVER. Our method provides a new detection framework for small-size objects, with potential applications in various fields.

FBrake, a Method to Simulate the Brake Efficiency of Laden Light Passenger Vehicles in PTIs While Measuring the Braking Forces of Their Unladen Configurations

This study introduces fBrake, a novel simulation method now designed for use in periodic technical inspections of M1 and N1 vehicle categories, addressing challenges posed by Directive 2014/45/EU. The directive mandates that braking efficiency must be measured relative to the vehicle’s maximum mass, which often results in underperformance during inspections due to vehicles typically being unladen. This discrepancy arises because the maximum braking forces are proportional to the vertical load on the wheels, causing empty vehicles to lock their wheels prematurely compared to laden ones. fBrake simulates the braking forces of unladen vehicles to reflect a laden state by employing an optimal brake-force distribution curve that aligns with the vehicle’s inherent braking behavior, whether through proportioning valves or through electronic brake distribution systems in anti-lock-braking-system-equipped vehicles. Our methodology, previously applied to heavy vehicles, involved extensive experimentation with a roller brake tester, comparing the actual braking performances of dozens of vehicles to those of their simulated counterparts using fBrake. The results demonstrate that fBrake reliably replicates the braking efficiency of laden vehicles, validating its use as an accurate and effective tool for braking system assessments in periodic inspections, irrespective of the vehicle’s load condition during the test. This approach ensures compliance with regulatory requirements while enhancing the reliability and safety of vehicle inspections.

Transformation and Development of New Energy Vehicle Industry Based on Social Policy Inquiry

The new energy vehicle industry has drawn the interest of international attention with the growing global demand for sustainability and reduction of carbon footprint. The growth of the New Energy Vehicle (NEV) industry is profoundly influenced by global policies aimed at driving technological innovation, meeting environmental requirements, and facilitating economic transformation. Multiple stakeholders, from policymakers to companies, from consumers to technology developers, are working together to drive the rapid growth of this industry. By analyzing research and practices from various nations and areas, insights can be gained on strategies for successfully advancing the growth of the new energy vehicle sector. From a social point of view more environmentally friendly, reducing the pollution of the traditional class of oil vehicles on the environment; from the consumer point of view, due to the fierce competition in the growth of the automotive new energy sector, the businessman will provide more tempting products to attract consumers, to promote the consumer to pick the price is more suitable for the quality of the product better, and further increase customer satisfaction. The future development of the new energy vehicle industry relies on progress on multiple fronts: policy, technology, and market. The healthy development and environmental sustainability of the industry can be more effectively promoted through international cooperation, technological innovation, and a deeper understanding of consumer needs. Policymakers need to take these factors into account to ensure that policies are both relevant to current market and technological realities and adaptable to future changes.

Estimates of Air Pollutant Emissions by Vehicle Fleet in the State of Pará, Brazil: Impacts and Trends

Objective: aims to estimate emissions of the main atmospheric pollutants from vehicular sources in the state of Pará, to continue the environmental agenda of monitoring air quality. Theoretical Framework: The growing global concern about air quality in urban areas is driven by the environmental and public health impacts caused by air pollution. The concentration of pollutants from vehicle sources represents one of the main challenges, especially in densely populated regions such as the capital and other major cities in the state of Pará, where the automotive fleet has increased significantly. Method: The methodology used combined "bottom-up" and "top-down" methods to calculate emissions of carbon monoxide (CO), nitrogen oxides (NO) and total hydrocarbons (HC) for different vehicle categories and fuel types. Data on licensed vehicle fleets and fuel consumption obtained from DETRAN-PA and ANP were used. Results and Discussion: The results revealed that gasoline-powered vehicles are responsible for the majority of CO and HC emissions, while diesel vehicles emit the largest amounts of NO into the atmosphere. Motorcycles were identified as the largest source of emissions due to their larger number in the circulating fleet. Our results also highlight the importance of assessing and monitoring vehicle emissions to develop effective air pollution control strategies. Research Implications: The information obtained in this study is unprecedented for the study area, and can therefore support the formulation of environmental policies aimed at reducing emissions of atmospheric pollutants in the state of Pará, thus promoting an improvement in air quality and public health. Originality/Value: This paper makes an unprecedented contribution to national and international literature on environmental pollution in large urban centers. On the eve of the largest climate and environment conference on the planet (COP30) to be held in Pará, Amazonas, this research joins efforts in relation to the monitoring, forecasting and mitigation of air pollution, considered one of the main environmental problems of today.

Trajectory optimization and stability control for a novel unmanned intelligent wheel-legged vehicles on unstructured terrains

The trend of intelligent vehicles is currently expanding, and a new class of unmanned intelligent vehicles known as wheel-legged vehicles (WLVs) is emerging. WLVs excel in transportation on unstructured terrain by offering a unique combination of the efficiency of wheels on flat ground and the versatility of legs to tackle obstacles. To enhance the locomotion performance of WLVs on unstructured terrains, this paper presents a novel framework for improving their locomotion through nonlinear programming-based (NLP) trajectory optimization and stability control. The framework optimizes the vehicle’s body and wheel positioning while incorporating terrain information and employs a linear rigid body dynamic model for efficient motion planning. The stability control framework combines feedforward control using ground reaction forces with feedback control through joint PD control and utilizes model predictive control (MPC) to adjust the wheel slip ratio to prevent slip on steep slopes. Experimental validation on the real vehicle with torque-controlled wheels demonstrated the capability of driving over a 1 m height with a 30°slope at an average speed of 0.7 m/s and a maximum speed of 1.03 m/s. Our approach also enables the WLV to overcome obstacles, such as inclines, while dynamically negotiating these challenging terrains.

Study the Level of Service to Roadways which Leading to Basrah University

The building of Basrah University located in a rural area at Gramat Ali, its consist of seven colleges and very high number of students, lecturers ,and employers. Therefore, the intersections which leading to university building suffer congestion at morning peak hour (8-9) A.M and evening peak hour (2-3) P.M. In this study we collected the data in intersections which leading to university building at peak hour at same time, then we had analysis this data by using the highway capacity manual (HCM) and program (HCS). Also, we collected that data in University entries with classification of vehicles according to types and number of passengers at peak hour. Then we studied the number of proposals to riddance the congestion, such as the effect of vehicles prohibition which one and two passenger to entry in University at peak hour. This proposal will riddance of the congestion at University entries and intersections.

Simplified method to create vehicle sound model for road traffic noise simulation-By using sound limit value of noise regulation

Road traffic noise prediction model is used to evaluate and analyze current situation and predict future prospect for road traffic noise. The current models such as CNOSSOS-EU and ASJ-RTN (Acoustical Society of Japan-Road traffic noise prediction model) define the fixed vehicle sound power level formulas. There are vehicle sound formulas for limited vehicle categories, but they are not enough to analyze influence of emitted sound from the vehicles in detail. Looking at an actual road, there are several vehicle categories and old vehicles which mean different regulation stages (e.g. sound limit). Since it is not easy to update the vehicle sound formulas with current methodology such as ISO11819-1 (Measurement of the influence of road surfaces on traffic noise -Part1 Statistical Pass-by method), we have proposed the method for creating vehicle sound formulas with less measurements in proving ground. Furthermore, when test vehicles are difficult to obtain as in the case of vehicles that comply with older regulations, we would like to propose an approximated method for practical use from sound limit in vehicle regulation and database in this paper. And the proposed method is validated comparing with the precise model.

Impact of road noise components (rolling/propulsion) by vehicle categories (light-duty, heavy-duty and two-wheeled vehicles) on population's noise exposure and associated adverse health effects

Exposure to road traffic noise can lead to various adverse health effects, ranging from annoyance and night-time sleep disturbances to ischemic heart diseases. The aim of this paper is to identify the contribution of the road noise components (rolling/propulsion) for each vehicle category (light-duty, heavy-duty and two-wheeled vehicles) on population's noise exposure and the associated adverse health effects. This analysis focuses on the case study of Lyon, France. The input data align with the requirements of the 4th deadline of Directive 2002/49/CE and adhere to the CNOSSOS calculation standard. The results reveal that the rolling noise component of light-duty vehicles is predominant due to vehicle speed and traffic volume. It is the sole noise component (i) significantly exposing a portion of the population to Lden > 68 dBA, noise limit set by the French decree of April 4, 2006, and (ii) where over half of the population is exposed to Lden > 53 dBA, recommended threshold by the World Health Organization (2018). By using transfer functions established between noise components for each vehicle category, and population's noise exposure, noise source-abatement solutions can be quantified in terms of their impact on population's noise exposure, thus determining the reduction in health impacts.

The problem of obtaining vehicles speed input for road traffic noise microscopic models: a comparison between field measurements and a stochastic approach

Mitigating environmental noise harmful effects is a relevant goal in the EU "Zero Pollution Action Plan", which targets a 30% reduction in chronic annoyance from transportation noise by 2030. Since road traffic is a major noise source in urban and non-urban settings, much effort has been invested in analyzing and modeling it. To assess noise levels, numerous Road Traffic Noise Models (RTNMs) exist, each requiring multiple inputs like vehicle flows, percentage of heavy vehicles, and average speed. Specifically, when dealing with microscopic models, the single vehicle speeds are needed. To obtain them, on-site measurements offer accurate data but are time consuming and expensive. This paper explores two approaches: an alternative measurement approach, based on video processing and object detection tools, and a stochastic approach, which randomly assigns a speed to each vehicle from a distribution curve, depending on traffic conditions and vehicle category. These methodologies are used to provide input data to a road traffic noise microscopic model, whose results are compared with field measured data. This comparison will allow to get useful insights on the performances of the proposed techniques in providing reliable inputs and their potential applications in different scenarios where single vehicle measured speeds are not available.

Концепция использования современных технических средств при проведении обследования помещений, ограничивающего конституционное право человека и гражданина на неприкосновенность жилища

The article provides a systematic analysis of current legislation, generally accepted judicial practice and various doctrinal approaches to the legal regulation of public inspection of premises, which restricts the constitutional right of a person and citizen to inviolability of the home, and suggests ways to improve current legislation aimed at eliminating gaps in the legal regulation of the procedure for conducting this operational search event. As a result of the study, special powers of employees of operational units were noted when drawing up a protocol for the inspection of a home and using video and audio recordings, film and photography, as well as other technical and other means during this operational search event. The paper describes the legal regime and classification of unmanned aerial vehicles depending on the maximum take-off weight and type of ownership. Taking into account modern scientific and technical achievements, it is proposed to use a quadcopter using the following innovative methods when conducting a housing survey: video shooting, panoramic, nodal and detailed photography, “online broadcast”, three-dimensional modeling, modeling using artificial intelligence, digital protocol compilation and methods: photo and (or) video shooting using infrared and ultraviolet detectors, video shooting using gas analysis detectors, X-ray photography. The use of modern technical means in conducting a home survey will ensure objectivity, increase efficiency and guarantee high efficiency of this operational investigative measures.

Methods to estimate efficiency of Low Noise Pavement for specific vehicle categories

Low Noise Pavements (LNPs) are efficient in reducing the rolling noise component of road-tire noise generation mechanisms. Open (porous) asphalt can be also efficient in reducing the propulsion noise component. As rolling/propulsion noise ratio varies with vehicles' category, the mitigation effect of a LNP varies between categories as confirmed by experimental data gathered in the LIFE NEREIDE. A proper model needs to predict and simulate the effects of LNPs in real conditions in the implementation sites. At present, LNPs can be modeled according to CNOSSOS-EU methodology using available reference surfaces, or by tuning user-defined surfaces. The paper discusses the limitations of mapping methods in delivering reliable values for specific vehicle categories, even when using ad-hoc tuned surfaces. The results are based on case studies in which modeling results are compared to experimental data. The study found that differences in measurements are due to the way the source model is conceived and cannot be completely eliminated.

CNOSSOS-FR: Are current road traffic noise emission coefficients still representative of today's vehicle fleet and roads?

CNOSSOS-EU is the common noise prediction method for producing strategic noise maps in Europe, according to Directive 2002/49/CE. Its road vehicle noise emission model involves a propulsion noise component and a rolling noise component, with coefficients specific to each vehicle category. As currently applied in France, it relies on the original EU coefficients published in the 2015 CNOSSOS-EU version and the rolling noise contribution is detailed in road surface classes in agreement with the national NMPB2008 method. As provided in the European method, corrective coefficients in frequency bands were determined per road surface class. The objective was to minimise the overall difference between NMPB2008 and CNOSSOS-FR emission predictions under similar conditions, the former method being previously used for periodic mapping. Now considering the later release of new EU reference coefficients and the difficulty of making the two methods involving different expressions and assumptions coincide, as well as questioning the need to update emission values resulting from measurements of the vehicle fleet taken 20-25 years ago, the study MINOOS is being carried out to assess the noise emission of the current vehicle fleet and reconsider French prediction coefficients accordingly. The paper describes the procedure and presents the first results.

Learning Observer Based Fault Estimation for a Class of Unmanned Marine Vehicles: The Switched System Approach

Learning Observer Based Fault Estimation for a Class of Unmanned Marine Vehicles: The Switched System Approach

KLASIFIKASI DAN VOLUME KENDARAAN BINA MARGA DARI REKAMAN VIDEO LALULINTAS DENGAN METODE ARTIFICIAL INTELLIGENCE - YOLO V8 PADA JALAN TOL JAKARTA-BOGOR DAN JAKARTA-TANGERANG

The software helps solve things faster, e.g., Epanet software, Pipe Flow Expert, WaterCAD, SAP2000 planning software, etc. Likewise, in the field of calculating traffic volume with the development of artificial intelligence technology, using video recording, volume, and speed calculations can be done easily. At this time, the classification of traffic counting is only up to 3 (three) types of vehicles classified by YOLO V8, so in this research, the classification is added to be 6 (six) types of vehicles, namely mini cars, mobile passengers, small buses, big buses, small trucks, and large trucks, according to vehicle classification established in Indonesia. The research method is done by recording the traffic on the road and then video recording with the help of the software developed with the Python program so that the volume of the vehicle according to its classification is obtained. The result obtained from this research is compared to the manual traffic counting. The comparison between vehicle counting by the software and manual counting is mini-car 97.02%, passenger car 99.49%, small bus 96.94%, large bus 97.34%, small truck 93.21%, and large truck 97.18%.

Vehicle Class Prediction at Toll Gate Using Deep Learning

In the era of digitalization and automation, efficiency in the traffic management system at toll gates is very important. One of the efforts to improve this efficiency is to develop an automatic vehicle class detection system using deep learning technology, especially Convolutional Neural Network (CNN). This research aims to design and implement a CNN model that can identify and classify the types of vehicles passing through toll gates. The model development process includes collecting and annotating vehicle image data, data pre-processing, and CNN model training and testing. The evaluation results show that the developed model can achieve an accuracy of about 96% in detecting vehicle classes, so it can be integrated with the toll gate system to increase the speed and accuracy in the vehicle classification process. Thus, this solution is expected to reduce the waiting time of toll users and improve operational efficiency.

Vehicle Detection and Counting for Traffic Congestion Estimation Using YOLOv5 and DeepSORT in Smart Traffic Light Application

The escalating number of vehicles on the roads has exacerbated trafficcongestion, presenting a significant and inevitable challenge for roadusers.Trafficcongestionnotonlyincreasesthetraveltimeofroadusers;it also causes air pollution since more vehicles are stuck on theroad. To address this issue, deep learning algorithms, including YOLOv5and DeepSORT, have been leveraged to enable vehicle detection and estimate the number of vehicles through image processing. This study focuses on training a custom YOLOv5 dataset of vehicles and compares its performance with a pretrained YOLOv5 dataset in terms of feasibility for detecting and estimating the number of vehicles. In the proposed approach, YOLOv5 is employed to detect vehicles in video streams, DeepSORT is used for vehicle tracking and counting as they pass through the detection zone, and the number of vehicles in each lane is estimated with the aid of Supervision. The custom dataset's validation demonstrates promising results, with the precision curve indicating convergence for all classes at 97.4% precision and an 87% accuracy in predicting vehicle classes. Additionally, the precision-recall curve assesses the ability to detect individual vehicle categories, such as bus, car, motorcycle, and truck, with accuracies of 68.2%, 95.2%, 79.9%, and 70.1%, respectively. Furthermore, the overall accuracy for detecting all vehicle classes combined is 78.3%. The F1 curve indicates that the system achieves a confidence level of 30.9% F1 score, ensuring 78% confidence in accurately predicting all classes. Both the pretrained and custom datasets exhibit similar accuracy in counting the total number of vehicles passing through the detection zone as well as estimating the number of vehicles in each lane. However, it is evident that the custom dataset's performance can be further improved by incorporating more extensive and diverse datasets during the training process to enhance the accuracy of vehicle detection and estimation. In conclusion, the integration of deep learning algorithms, specificallyYOLOv5 and DeepSORT, offers a promising solution for addressing traffic congestion byefficiently detecting and estimating vehicle numbers, which allow traffic systems to identify which lane is congested and prioritise ensuring the congested lane has a smooth traffic flow. Continued research with larger datasets can lead to further advancements and refined results in the field of traffic management and optimization.