Real-time Road Research Articles

Research on Pavement Defect Detection Algorithm Based on SEM-YOLOv8n

Automated detection and identification of pavement distresses is essential for timely pavement repair. Subtle pavement defects and multiple defects detection is a challenging task under complex background. With the deepening of the deep learning network, some subtle features tend to disappear and are more difficult to detect under the influence of the complex background. To solve the above problems, this paper proposes the SEM-YOLOv8n pavement defect detection algorithm. Firstly, SPD-Conv is used to replace the traditional convolution, which is conducive to retaining more defect detail information in the image and improving the detection ability of subtle defects; then an efficient multi-scale attention mechanism is added to the fusion network, so that the network suppresses the background information and focuses more on the defect information. Finally, MPDIoU is introduced as a loss function, which optimizes the minimum perpendicular distance between the predicted bounding box and the real bounding box and improves the localization ability, thus improving the accuracy of the network. Finally, the effectiveness of the proposed network is verified on the IRRDD dataset, and the results show that the method achieves 91.9% (Precision), 91.3% (Recall), and 71.3% (mAP) for the classification and detection of road multi-scale minor defects, which meets the demand of real-time road defect detection.

Real-Time Road Safety: A Web and Mobile Application Approach to Mitigating Animal-Vehicle Conflicts in Sri Lanka

This research paper presents a pioneering web and mobile application aimed at mitigating the rising issue of animal-vehicle collisions in Sri Lanka, a concern that poses significant risks to both human safety and wildlife preservation. The application leverages a combination of knowledge-based case studies and crowd- sourcing techniques to enhance real-time road safety by identifying animal habitats and behaviors in proximity to users' locations. Through the web platform, users can input data on observed wildlife interactions and animal- vehicle conflicts, contributing to a comprehensive database for statistical analysis and predictive modelling. This information is used to generate real-time alerts and feedback, enabling drivers to take precautionary measures. The mobile application, utilizing geofencing and geotargeting technologies, provides real-time alerts, facilitates user feedback, and fosters community participation in ensuring safer roads. In addition to mitigating accidents, the application serves as an educational tool, offering guidelines and raising public awareness about the importance of road safety in areas rich in wildlife. By fostering a collaborative effort among users, this solution aims to reduce the frequency of animal-vehicle collisions, thereby promoting safer roads for both humans and animals in Sri Lanka.

Auto-GeRo: An IOT based Geo Spatial Model for Real-Time Road Condition Detection

Auto-GeRo: An IOT based Geo Spatial Model for Real-Time Road Condition Detection

Transferability of real-time EVT safety models: an investigation using autonomous vehicles data

Recent research has shown that data from autonomous vehicles (AVs) enables proactive, real-time road safety management. However, with low AV market penetration, it's crucial to assess the transferability of AV-based safety models to other locations. This study transfers real-time safety models from a data-rich to a data-scarce environment using vehicular conflict data collected by AVs. Multi-site Bayesian hierarchical Extreme Value Theory models are developed in two cities, with model transfer evaluated using informative priors and parameter recalibration. Both approaches produce reliable results, though the transferability index favors the informed prior method, while predictive deviance shows similar performance for both. Model sensitivity to temporal data limitations is assessed, revealing that the recalibration approach is more sensitive to block size, whereas the informed prior method remains stable. A novel method mimicking data limitations shows transferred models perform well with data utilizations above 20%, promoting transferability to mitigate data inadequacy.

Redefining Real-Time Road Quality Analysis With Vision Transformers on Edge Devices

Redefining Real-Time Road Quality Analysis With Vision Transformers on Edge Devices

Automatic target detection in vehicle images based on YOLOv10 deep learning algorithm

In this paper, we explore the effect of its application in vehicle image detection based on the latest YOLOv10 model.YOLOv10, as the latest version in the YOLO series, inherits and optimises the advantages of the previous generations of models, aiming to provide higher detection accuracy and faster inference speed. In our experiments, we firstly divided the dataset reasonably and trained the YOLOv10 model with data from the training set. By analysing the Precision-Recall curve, we found that the area below the PR curve is larger, which indicates that the AUC value of the model is close to 1, thus reflecting the superiority of the model in classification performance. In addition, from the F1-confidence curve, the model is still able to maintain good classification results at higher confidence levels and maintains high F1 scores in most confidence intervals, which further validates the reliability of the model. Test results show that the trained YOLOv10 model is able to accurately recognise vehicles and effectively predict vehicle types. This enables the model to monitor the vehicle conditions on the road in real time and achieve accurate detection and classification of road traffic targets. Therefore, this paper demonstrates the powerful ability of YOLOv10 in the field of vehicle detection, which provides an important reference for the development of future intelligent transport systems. Overall, this study not only verifies the effectiveness of YOLOv10 in practical applications, but also lays the foundation for further improvement of target detection technology.

Efficient Method of Road Outlier Recognition Using Deep Learning Coupled with Data Augmentation Approach

In recent years, Closed-circuit Television (CCTV) cameras have been playing a vital role in the surveillance of both public and private areas. The primary objective of surveillance is to monitor human behavior and road conditions. In the real-world situation, detecting, and recognizing abnormal activities poses significant challenges due to the densely crowded environment and the complex nature of transportation systems. These factors make it difficult to automatically identify various anomalies that occur while traveling, leading to emergencies, and endangering human life and property. In response to these challenges, contemporary Road Outlier Recognition Surveillance Systems (RORSS) have been deployed to monitor and mitigate abnormalities on streets, highways, and roads. This research specifically investigated conventional as well as human-centered road anomalies encompassing snatching, accidents, fighting, and car fires. The proposed Real-time Road Outlier Recognition methodology is formulated as a classification task, involving the analysis and processing of real-time CCTV videos. The proposed study investigated different types of Convolutional Neural Network (CNN) pre-trained models with the Data Augmentation (DA) approach to address the frame variance problem in real-time videos. Furthermore, we introduced a benchmark real-world Road Outlier Dataset (ROD) containing roads, streets, and highways videos and images that presented different road anomalies performed by vehicles and humans. Experimentation using different pre-trained CNN models i.e., VGG19, InceptionV3, ResNet50, and DenseNet201 with a data augmentation approach has been performed on the ROD dataset. The experimental results demonstrated that the InceptionV3 model performed best with the augmentation approach as compared to other deep learning models that achieved the best accuracy of 98.80% in recognizing road anomalies

Multi-objective optimization of PI controller for BLDC motor speed control and energy saving in Electric Vehicles: A constrained swarm-based approach

Proportional Integral Derivative (PID) controllers are widely employed in industrial applications due to their effectiveness, simplicity, and versatility. Within the automotive sector, one notable application of the Proportional Integral (PI) controller is in the realm of electric motors, particularly the Brushless Direct Current (BLDC) motors powering Electric Vehicles (EVs). The precision of speed control over BLDC motors, renowned for their efficiency and reliability, is paramount for optimizing vehicle performance and ensuring energy efficiency. PIDs must be tuned each time they are used in an application to optimize performance. A well-known empirical hit and trial tuning strategy is the Ziegler–Nichols method, which results in sub-optimal performance in the presence of locally optimal solutions of large-dimensional search spaces. Existing Artificial Intelligence methods tailored for PID controller tuning often focus on singular optimization aspects, neglecting potential performance gains achievable through multi-objective considerations. Furthermore, prevalent swarm-based PID optimization methods typically initialize the population randomly without imposing any bounds on input parameters, thus rendering them susceptible to inconsistent, unreliable, sub-optimal solutions in unconstrained search environments. In our research endeavors, we mathematically model the multi-objective optimization of BLDC motor speed control, energy usage, and efficiency using constrained Differential Evolution and Particle Swarm Optimized PID controllers. The Ziegler–Nichols tuning method is used to ascertain the initial PI parameters and define bounds within the input space. Our methodology addresses a critical gap by integrating real-time road gradient data into the EV control system, enhancing precision and minimizing energy consumption, especially during challenging road conditions. The proposed approach significantly improves motor speed control and energy efficiency in EVs. With our proposed swam-based optimization over 30 iterations, MSE decreased by approximately 95.4% (from 3.8834 to 0.1809), while energy consumption reduced by about 3.1% (from 1.015 kWh to 0.984 kWh). These results highlight the effectiveness of our method in enhancing both speed control precision and energy efficiency. In addition, code related data is available at Github.

Real-Time Road Damage Detection and Infrastructure Evaluation Leveraging Unmanned Aerial Vehicles and Tiny Machine Learning

Real-Time Road Damage Detection and Infrastructure Evaluation Leveraging Unmanned Aerial Vehicles and Tiny Machine Learning

BSNet: A bilateral real-time semantic segmentation network based on multi-scale receptive fields

The rise of autonomous driving and mobile robots has drawn attention to real-time road scene segmentation. However, category confusion, incomplete segmentation and inaccurate detail contours are common issues encountered in road traffic scene segmentation tasks. To tackle these road segmentation challenges, we proposes a bilateral real-time semantic segmentation network based on multi-scale receptive fields(BSNet). To enhance the segmentation accuracy of detail contours, an auxiliary detail detection module is introduced in the spatial branch. In the semantic branch, strip pooling and channel attention are added to the short-term dense concatenate module to enrich multi-scale receptive fields and strengthen channel features, addressing issues such as incomplete segmentation and category confusion. The final design of the complementary dual-guided fusion module improves feature fusion effectiveness. Experimental results on two benchmark datasets demonstrate that BSNet significantly improves these road segmentation difficulties and achieves competitive results in both speed and accuracy compared to the current state-of-the-art methods.

Designing of self balancing amplitude modulated five level inverter for reducing voltage stress gradients on converter switches for electric vehicles

The research article endeavors to explore the efficient multi-quadrant operations in Electric Vehicles (EV), which emphasizing among the renewable sources and Energy Storage Management Systems (ESMS). This coordination is facilitated through the latest advancements in converter circuits, aiming to enhance efficiency and performance in EV operations. The primary converting circuit, or inverter, is crucial in optimizing the utilization of energy sourced from renewable sources. The coordination among these systems along with the converter, produces harmonics, due to the shifting of operating modes in EV motors in quick sessions because of dynamic real-time road conditions. This research paper presents the design and implementation of Self Balancing Modulated Five-Level Inverter along with a powerful modulation technique and coordinated with the ESMS of electric car application. The proposed 5-Level inverter shows the reduction in torque distortions in an electric vehicle drives over conventional two-stage boost inverter (2SBI) circuits. The designed Five-Level Inverter (FLI) circuit characterized by reduced switching operations, shows a significant reduction in voltage stress gradients on switches and achieves self-balancing of voltages with reduced operating components. The developed three-phase boost single circuit named as SBAM-FLI is connected to the grid integrated renewable energy application and the system is tested at constant and variable torque conditions. A comparative analysis is conducted between the operation of self-balancing modulated FLI with 2SBI based on the obtained results and power loss analysis is performed for the performance evolution in EV applications. The proposed topology is adaptable to both the Trapezoidal and Sinusoidal Back EMF drive mechanisms, which shows significant improvement over conventional inverters. The experimental validation of proposed self-balancing modulated FLI design is presented in this research. The experimental setup was realized using Xilinx block sets and seamlessly integrated with the Basys-3 Artix-FPGA board, consequently, both the experimental and simulation outcomes of the proposed converter topology were validated and provides substantial impact on EV applications.

Edge intelligence empowered delivery route planning for handling changes in uncertain supply chain environment

Traditional delivery route planning faces challenges in reducing logistics costs and improving customer satisfaction with growing customer demand and complex road traffic, especially in uncertain supply chain environment. To address these challenges, we introduce an innovative two-phase delivery route planning method integrating edge intelligence technology. The novelty of our approach lies in utilizing edge computing devices to monitor real-time changes in road conditions and dynamically adjust delivery routes, thereby providing an effective solution for efficient and flexible logistics. Initially, we construct a mixed-integer programming model that minimizes the total cost under constraints such as customer destinations and time windows. Subsequently, in the cloud-edge collaborative mode, edge computing devices are utilized to collect real-time road conditions and transmit it to the cloud server. The cloud server comprehensively considers customer demand and road condition changes and employs adaptive genetic algorithms and A-star algorithms to adjust the delivery routes dynamically. Finally, comprehensive experiments are conducted to validate the effectiveness of our method. The results demonstrate that our approach can promptly respond to changes in customer demands and road conditions and flexibly plan the optimal delivery routes, thereby significantly reducing overall costs and enhancing customer satisfaction.

An optimization model for just-in-time (JIT) delivery of precast components considering 3D loading constraints, real-time road conditions and assembly time

PurposeAlthough extensive research has been conducted on precast production, irregular component loading constraints have received little attention, resulting in limitations for transportation cost optimization. Traditional irregular component loading methods are based on past performance, which frequently wastes vehicle space. Additionally, real-time road conditions, precast component assembly times, and delivery vehicle waiting times due to equipment constraints at the construction site affect transportation time and overall transportation costs. Therefore, this paper aims to provide an optimization model for Just-In-Time (JIT) delivery of precast components considering 3D loading constraints, real-time road conditions and assembly time.Design/methodology/approachIn order to propose a JIT (just-in-time) delivery optimization model, the effects of the sizes of irregular precast components, the assembly time, and the loading methods are considered in the 3D loading constraint model. In addition, for JIT delivery, incorporating real-time road conditions in the transportation process is essential to mitigate delays in the delivery of precast components. The 3D precast component loading problem is solved by using a hybrid genetic algorithm which mixes the genetic algorithm and the simulated annealing algorithm.FindingsA real case study was used to validate the JIT delivery optimization model. The results indicated this study contributes to the optimization of strategies for loading irregular precast components and the reduction of transportation costs by 5.38%.Originality/valueThis study establishes a JIT delivery optimization model with the aim of reducing transportation costs by considering 3D loading constraints, real-time road conditions and assembly time. The irregular precast component is simplified into 3D bounding box and loaded with three-space division heuristic packing algorithm. In addition, the hybrid algorithm mixing the genetic algorithm and the simulated annealing algorithm is to solve the 3D container loading problem, which provides both global search capability and the ability to perform local searching. The JIT delivery optimization model can provide decision-makers with a more comprehensive and economical strategy for loading and transporting irregular precast components.

Exploring particle concentrations and inside-to-outside ratios in vehicles: A real-time road test study

In transportation microenvironments, humans exposed to particulate matter (PM) inside vehicles can experience higher levels of daily exposure. To make inside-vehicle PM exposure measurements more feasible and easy under real driving conditions, and to quantify the relationship between the concentrations and influencing factors, we assessed PM1, PM2.5, and PM10. levels. Additionally, we collected key influencing factors to develop predictive models. The measurements of PM1, PM2.5, and PM10 concentrations showed that the ventilation setting was a significant influencing factor. The concentrations decreased significantly under the recirculation setting (RC) compared to the outside air setting (OA). The inside-to-outside (I/O) ratios of PM were 1.69 to 1.93-fold higher than those of RC under OA conditions. However, a substantial reduction in the I/O ratios was observed when RC was employed. Although both the concentrations and I/O ratios exhibited significant differences, they demonstrated strong potential relationships. PM2.5 I/O ratios accounted for over 85 % of the variation in the PM1 and PM10 I/O ratios. The developed models for the I/O ratios of PM accounted for >40 and 60 % of the variation in the measured I/O ratios for RC and OA, respectively. We used the vehicle age, vehicle interior volume, speed, cabin temperature, cabin humidity, and their higher-order terms as predictive variables. It is important to note that the influential predictive feature importance differed under RC and OA, and considering the vehicle characteristics between vehicles of the same type may be necessary when using RC. Overall, these findings indicate that the inside-vehicle PM exposure can be measured more easily under real driving conditions by considering the key influencing factors and utilizing the developed predictive models.

MED-YOLOv8s: a new real-time road crack, pothole, and patch detection model

MED-YOLOv8s: a new real-time road crack, pothole, and patch detection model

Monitoring Rubbish on Roads in Real Time by Deep Neural Network

Monitoring and managing litter on the roadways is important not only for preserving the cleanliness and esthetic appeal of our cities but also for safeguarding the overall health and sanitary environment of citizens. With the development of artificial intelligence, it is now possible to design algorithms capable of autonomously assessing the sanitation status of roadways. In this paper, we propose a method for detecting garbage on roads as our solution to tackle the issue. The proposed model is a deep learning model with an attention mechanism introduced to detect garbage in real-time scenarios. Furthermore, the object recognition capability of the model is enhanced through transfer learning to mitigate the influence of irrelevant content during training. The refined network can detect rubbish on roads with higher efficiency than the existing models. Also, compared with other methods, it consumes less amount of data for training. The experimental results demonstrate the efficiency of our proposed model, with a performance boost of 7.62% over the state-of-the-art methods.

Investigating Driver Preferences for Traffic Information Using Digital Signage and Road Surface Holograms

This study investigates road users' information preferences to enhance traffic information provision strategies. Traditional road signs primarily convey static information like speed limits and restricted areas. However, smart street poles now offer the capability to collect real-time road conditions and deliver dynamic traffic updates promptly. Leveraging these smart poles allows for more effective and tailored information delivery strategies. The research examines three types of media for information dissemination: digital signage, road surface hologram, and a combination of both. Additionally, it surveys road users' preferences for information delivery methods, including text, graphics, or a mixture, based on the type of information. Conjoint analysis of 220 responses reveals essential attributes influencing users when they receive information. The findings indicate that a majority of road users prefer a blend of digital signage and road surface holograms, with a notable preference among younger individuals for digital signage. For road surface holograms, the favored approach involves a mix of text and graphics. This study lays the foundation for user-centric information strategies, offering insights into tailoring approaches based on the type of information and user preferences.

Frequency Based Supervisory Power Management Approach for Battery-Supercapacitor Electric Vehicle Considering Road Slope

This article proposes a real time strategy based on frequency separation for energy management of battery/Supercapacitor (SC) based Hybrid Energy Storage System (HESS) in Electric Vehicle (EV). The performance and life cycle of battery are important parameters in determining the efficiency of EVs. With the aim of improving the effectiveness of battery in battery/SC based EVs, a supervisory Symlets wavelet based Energy Management System (EMS) using Fuzzy Controller (FC) is proposed which considers real time road slope and vehicle condition. Actual road grade values are computed from Contour Positioning System (CPS) using Google Earth software. Symlets wavelet control technique is utilised to separate the high frequency transients from the required load power. After separation of high frequency components, a FC is developed to provide an additional reference power for battery and SC. Then, the high frequency transients along with additional power obtained from FC are allocated to SC and the rest is handled by the battery. Three road slope conditions viz. city tour, downhill and uphill and two drive cycles viz. NYCC and UDDS are applied to examine the capability of proposed EMS. The proposed Fuzzy Symlets Wavelet based controller is observed to improve the battery Ampere hour (Ah) throughput and Root Mean Square (RMS) value of battery current in comparison to only Symlets wavelet based controller for city tour, uphill and downhill journeys. The results indicate the robustness and superiority of the designed Fuzzy Symlets Wavelet based control technique as compared to Symlets Wavelet technique in handling practical conditions and improving vehicle performance.

A Real-Time Road Boundary Detection Approach in Surface Mine Based on Meta Random Forest

A Real-Time Road Boundary Detection Approach in Surface Mine Based on Meta Random Forest

Real-time Road Surface LiDAR SLAM based on Dead-Reckoning Assisted Registration

Real-time Road Surface LiDAR SLAM based on Dead-Reckoning Assisted Registration