Sensors

Research

17 pages, 2525 KB

Open AccessArticle

Intelligent Compaction System for Soil-Rock Mixture Subgrades: Real-Time Moisture-CMV Fusion Control and Embedded Edge Computing

by Meisheng Shi, Shen Zuo, Jin Li, Junwei Bi, Qingluan Li and Menghan Zhang

Sensors 2025, 25(17), 5491; https://doi.org/10.3390/s25175491 - 3 Sep 2025

Viewed by 793

Abstract

The compaction quality of soil–rock mixture (SRM) subgrades critically influences infrastructure stability, but conventional settlement difference methods exhibit high spatial sampling bias (error > 15% in heterogeneous zones) and fail to characterize the overall compaction quality. These limitations lead to under-compaction (porosity > [...] Read more.

The compaction quality of soil–rock mixture (SRM) subgrades critically influences infrastructure stability, but conventional settlement difference methods exhibit high spatial sampling bias (error > 15% in heterogeneous zones) and fail to characterize the overall compaction quality. These limitations lead to under-compaction (porosity > 25%) or over-compaction (aggregate fragmentation rate > 40%), highlighting the need for real-time monitoring. This study develops an intelligent compaction system integrating (1) vibration acceleration sensors (PCB 356A16, ±50 g range) for compaction meter value (CMV) acquisition; (2) near-infrared (NIR) moisture meters (NDC CM710E, 1300–2500 nm wavelength) for real-time moisture monitoring (sampling rate 10 Hz); and (3) an embedded edge-computing module (NVIDIA Jetson Nano) for Python-based data fusion (FFT harmonic analysis + moisture correction) with 50 ms processing latency. Field validation on Linlin Expressway shows that the system meets JTG 3430-2020 standards, with the compaction qualification rate reaching 98% (vs. 82% for conventional methods) and 97.6% anomaly detection accuracy. This is the first system integrating NIR moisture correction (R² = 0.96 vs. oven-drying) with CMV harmonic analysis, reducing measurement error by 40% compared to conventional ICT (Bomag ECO Plus). It provides a digital solution for SRM subgrade quality control, enhancing construction efficiency and durability. Full article

(This article belongs to the Special Issue AI and Smart Sensors for Intelligent Transportation Systems)

► Show Figures

Figure 1

23 pages, 1466 KB

Open AccessArticle

TMU-Net: A Transformer-Based Multimodal Framework with Uncertainty Quantification for Driver Fatigue Detection

by Yaxin Zhang, Xuegang Xu, Yuetao Du and Ningchao Zhang

Sensors 2025, 25(17), 5364; https://doi.org/10.3390/s25175364 - 29 Aug 2025

Viewed by 539

Abstract

Driving fatigued is a prevalent issue frequently contributing to traffic accidents, prompting the development of automated fatigue detection methods based on various data sources, particularly reliable physiological signals. However, challenges in accuracy, robustness, and practicality persist, especially for cross-subject detection. Multimodal data fusion [...] Read more.

Driving fatigued is a prevalent issue frequently contributing to traffic accidents, prompting the development of automated fatigue detection methods based on various data sources, particularly reliable physiological signals. However, challenges in accuracy, robustness, and practicality persist, especially for cross-subject detection. Multimodal data fusion can enhance the effective estimation of driver fatigue. In this work, we leverage the advantages of multimodal signals to propose a novel Multimodal Attention Network (TMU-Net) for driver fatigue detection, achieving precise fatigue assessment by integrating electroencephalogram (EEG) and electrooculogram (EOG) signals. The core innovation of TMU-Net lies in its unimodal feature extraction module, which combines causal convolution, ConvSparseAttention, and Transformer encoders to effectively capture spatiotemporal features, and a multimodal fusion module that employs cross-modal attention and uncertainty-weighted gating to dynamically integrate complementary information. By incorporating uncertainty quantification, TMU-Net significantly enhances robustness to noise and individual variability. Experimental validation on the SEED-VIG dataset demonstrates TMU-Net’s superior performance stability across 23 subjects in cross-subject testing, effectively leveraging the complementary strengths of EEG (2 Hz full-band and five-band features) and EOG signals for high-precision fatigue detection. Furthermore, attention heatmap visualization reveals the dynamic interaction mechanisms between EEG and EOG signals, confirming the physiological rationality of TMU-Net’s feature fusion strategy. Practical challenges and future research directions for fatigue detection methods are also discussed. Full article

(This article belongs to the Special Issue AI and Smart Sensors for Intelligent Transportation Systems)

► Show Figures

Figure 1

27 pages, 5654 KB

Open AccessArticle

Intelligent Detection and Description of Foreign Object Debris on Airport Pavements via Enhanced YOLOv7 and GPT-Based Prompt Engineering

by Hanglin Cheng, Ruoxi Zhang, Ruiheng Zhang, Yihao Li, Yang Lei and Weiguang Zhang

Sensors 2025, 25(16), 5116; https://doi.org/10.3390/s25165116 - 18 Aug 2025

Viewed by 614

Abstract

Foreign Object Debris (FOD) on airport pavements poses a serious threat to aviation safety, making accurate detection and interpretable scene understanding crucial for operational risk management. This paper presents an integrated multi-modal framework that combines an enhanced YOLOv7-X detector, a cascaded YOLO-SAM segmentation [...] Read more.

Foreign Object Debris (FOD) on airport pavements poses a serious threat to aviation safety, making accurate detection and interpretable scene understanding crucial for operational risk management. This paper presents an integrated multi-modal framework that combines an enhanced YOLOv7-X detector, a cascaded YOLO-SAM segmentation module, and a structured prompt engineering mechanism to generate detailed semantic descriptions of detected FOD. Detection performance is improved through the integration of Coordinate Attention, Spatial–Depth Conversion (SPD-Conv), and a Gaussian Similarity IoU (GSIoU) loss, leading to a 3.9% gain in mAP@0.5 for small objects with only a 1.7% increase in inference latency. The YOLO-SAM cascade leverages high-quality masks to guide structured prompt generation, which incorporates spatial encoding, material attributes, and operational risk cues, resulting in a substantial improvement in description accuracy from 76.0% to 91.3%. Extensive experiments on a dataset of 12,000 real airport images demonstrate competitive detection and segmentation performance compared to recent CNN- and transformer-based baselines while achieving robust semantic generalization in challenging scenarios, such as complete darkness, low-light, high-glare nighttime conditions, and rainy weather. A runtime breakdown shows that the enhanced YOLOv7-X requires 40.2 ms per image, SAM segmentation takes 142.5 ms, structured prompt construction adds 23.5 ms, and BLIP-2 description generation requires 178.6 ms, resulting in an end-to-end latency of 384.8 ms per image. Although this does not meet strict real-time video requirements, it is suitable for semi-real-time or edge-assisted asynchronous deployment, where detection robustness and semantic interpretability are prioritized over ultra-low latency. The proposed framework offers a practical, deployable solution for airport FOD monitoring, combining high-precision detection with context-aware description generation to support intelligent runway inspection and maintenance decision-making. Full article

(This article belongs to the Special Issue AI and Smart Sensors for Intelligent Transportation Systems)

► Show Figures

Figure 1

14 pages, 899 KB

Open AccessArticle

Multi-Robot Path Planning for High-Density Parking Environments Considering Efficiency and Fairness

by Jinhyuk Lee and Woojin Chung

Sensors 2025, 25(14), 4342; https://doi.org/10.3390/s25144342 - 11 Jul 2025

Viewed by 578

Abstract

As parking congestion at airport parking lots intensifies, high-density parking (HDP) systems with multiple parking robots are gaining attention for improving operational efficiency. However, conventional multi-agent pathfinding (MAPF) methods primarily focus on overall efficiency improvement, often neglecting the priority of individual parking tasks. [...] Read more.

As parking congestion at airport parking lots intensifies, high-density parking (HDP) systems with multiple parking robots are gaining attention for improving operational efficiency. However, conventional multi-agent pathfinding (MAPF) methods primarily focus on overall efficiency improvement, often neglecting the priority of individual parking tasks. Additionally, these methods assume robots are ideal agents, resulting in physically infeasible paths for parking robots. We propose a multi-robot path planning approach that balances efficiency and priority. The proposed method improves priority-based search (PBS) by dynamically adjusting priorities, thereby ensuring both operational efficiency and priority of individual vehicles. A simulator replicating a real airport parking environment with 100 parking slots and parking robots under development was implemented to validate the approach. Real-world parking data from an airport was used as input, demonstrating that the proposed autonomous parking system can effectively handle peak-season parking demand. The proposed method achieves a throughput exceeding 41 vehicles per hour with appropriate weight value, meeting the peak-season demand while maintaining acceptable fairness. Our approach provides a practical foundation for establishing time-based parking operation strategies and estimating the number of robots recommended for a given parking scenario. Full article

(This article belongs to the Special Issue AI and Smart Sensors for Intelligent Transportation Systems)

► Show Figures

Figure 1

17 pages, 2155 KB

Open AccessArticle

The TDGL Module: A Fast Multi-Scale Vision Sensor Based on a Transformation Dilated Grouped Layer

by Leilei Xie, Fenghua Zhu and Zhixue Wang

Sensors 2025, 25(11), 3339; https://doi.org/10.3390/s25113339 - 26 May 2025

Viewed by 556

Abstract

Effectively capturing multi-scale object features is crucial for vision sensors used in road object detection tasks. Traditional spatial pyramid pooling methods fuse multi-scale feature information but lack adaptability in dynamically adjusting convolution operations based on their actual needs. This limitation prevents them from [...] Read more.

Effectively capturing multi-scale object features is crucial for vision sensors used in road object detection tasks. Traditional spatial pyramid pooling methods fuse multi-scale feature information but lack adaptability in dynamically adjusting convolution operations based on their actual needs. This limitation prevents them from fully utilizing spatial hierarchies and contextual information. To address this challenge, we propose a Transformation Dilated Grouped Layer (TDGL) module, a fast multi-scale vision sensor based on deep learning, designed to enhance both efficiency and accuracy in road target feature extraction networks. The TDGL is built upon the Global Layer Normalization Convolution (GLConv) unit, which mitigates internal covariate shift by introducing scaling and offset parameters, modifying dilation strategies, and employing grouped convolution. These improvements enable the network to distinguish features at different scales effectively while optimizing spatial information processing and reducing computational costs. To validate its effectiveness, we integrate the TDGL module into the backbone of several YOLO models, forming the TDGL Net feature extractor. The experimental results obtained on the BDD100K dataset show that the mAP of the TDGL net reaches 40.3% with around 3.1M parameters. The inference speed of the TDGL net after transformation optimization reaches 58 FPS, which meets the requirement for the real-time detection of road obstacle targets by autonomous vehicles. Full article

(This article belongs to the Special Issue AI and Smart Sensors for Intelligent Transportation Systems)

► Show Figures

Figure 1

22 pages, 1566 KB

Open AccessArticle

Opportunistic Allocation of Resources for Smart Metering Considering Fixed and Random Wireless Channels

by Christian Jara, Juan Inga and Esteban Inga

Sensors 2025, 25(8), 2570; https://doi.org/10.3390/s25082570 - 18 Apr 2025

Viewed by 644

Abstract

This paper presents an optimization model for wireless channel allocation in cellular networks, specifically designed for the transmission of smart meter (SM) data through a mobile virtual network operator (MVNO). The model efficiently allocates transmission channels, minimizing smart grid (SG) costs. The MVNO [...] Read more.

This paper presents an optimization model for wireless channel allocation in cellular networks, specifically designed for the transmission of smart meter (SM) data through a mobile virtual network operator (MVNO). The model efficiently allocates transmission channels, minimizing smart grid (SG) costs. The MVNO manages fixed and random channels through a shared access scheme, optimizing meter connectivity. Channel allocation is based on a Markovian approach and optimized through the Hungarian algorithm that minimizes the weight in a bipartite network between meters and channels. In addition, cumulative tokens are introduced that weight transmissions according to channel availability and network congestion. Simulations show that dynamic allocation in virtual networks improves transmission performance, contributing to sustainability and cost reduction in cellular networks. This study highlights the importance of inefficient resource management by cognitive mobile virtual network and cognitive radio virtual network operators (C-MVNOs), laying a solid foundation for future applications in intelligent networks. This work is motivated by the increasing demand for efficient and scalable data transmission in smart metering systems. The novelty lies in integrating cumulative tokens and a Markovian-based bipartite graph matching algorithm, which jointly optimize channel allocation and transmission reliability under heterogeneous wireless conditions. Full article

(This article belongs to the Special Issue AI and Smart Sensors for Intelligent Transportation Systems)

► Show Figures

Graphical abstract

21 pages, 2174 KB

Open AccessArticle

Deep Learning Ensemble Approach for Predicting Expected and Confidence Levels of Signal Phase and Timing Information at Actuated Traffic Signals

by Seifeldeen Eteifa, Amr Shafik, Hoda Eldardiry and Hesham A. Rakha

Sensors 2025, 25(6), 1664; https://doi.org/10.3390/s25061664 - 7 Mar 2025

Viewed by 2356

Abstract

Predicting Signal Phase and Timing (SPaT) information and confidence levels is needed to enhance Green Light Optimal Speed Advisory (GLOSA) and/or Eco-Cooperative Adaptive Cruise Control (Eco-CACC) systems. This study proposes an architecture based on transformer encoders to improve prediction performance. This architecture is [...] Read more.

Predicting Signal Phase and Timing (SPaT) information and confidence levels is needed to enhance Green Light Optimal Speed Advisory (GLOSA) and/or Eco-Cooperative Adaptive Cruise Control (Eco-CACC) systems. This study proposes an architecture based on transformer encoders to improve prediction performance. This architecture is combined with different deep learning methods, including Multilayer Perceptrons (MLP), Long-Short-Term Memory neural networks (LSTM), and Convolutional Long-Short-Term Memory neural networks (CNNLSTM) to form an ensemble of predictors. The ensemble is used to make data-driven predictions of SPaT information obtained from traffic signal controllers for six different intersections along the Gallows Road corridor in Virginia. The study outlines three primary tasks. Task one is predicting whether a phase would change within 20 s. Task two is predicting the exact change time within 20 s. Task three is assigning a confidence level to that prediction. The experiments show that the proposed transformer-based architecture outperforms all the previously used deep learning methods for the first two prediction tasks. Specifically, for the first task, the transformer encoder model provides an average accuracy of 96%. For task two, the transformer encoder models provided an average mean absolute error (MAE) of 1.49 s, compared to 1.63 s for other models. Consensus between models is shown to be a good leading indicator of confidence in ensemble predictions. The ensemble predictions with the highest level of consensus are within one second of the true value for 90.2% of the time as opposed to those with the lowest confidence level, which are within one second for only 68.4% of the time. Full article

(This article belongs to the Special Issue AI and Smart Sensors for Intelligent Transportation Systems)

► Show Figures

Figure 1

17 pages, 3343 KB

Open AccessArticle

Comparative Analysis of YOLO Series Algorithms for UAV-Based Highway Distress Inspection: Performance and Application Insights

by Ziyi Yang, Xin Lan and Hui Wang

Sensors 2025, 25(5), 1475; https://doi.org/10.3390/s25051475 - 27 Feb 2025

Cited by 10 | Viewed by 2193

Abstract

Established unmanned aerial vehicle (UAV) highway distress detection (HDD) faces the dual challenges of accuracy and efficiency, this paper conducted a comparative study on the application of the YOLO (You Only Look Once) series of algorithms in UAV-based HDD to provide a reference [...] Read more.

Established unmanned aerial vehicle (UAV) highway distress detection (HDD) faces the dual challenges of accuracy and efficiency, this paper conducted a comparative study on the application of the YOLO (You Only Look Once) series of algorithms in UAV-based HDD to provide a reference for the selection of models. YOLOv5-l and v9-c achieved the highest detection accuracy, with YOLOv5-l performing well in mean and classification detection precision and recall, while YOLOv9-c showed poor performance in these aspects. In terms of detection efficiency, YOLOv10-n, v7-t, and v11-n achieved the highest levels, while YOLOv5-n, v8-n, and v10-n had the smallest model sizes. Notably, YOLOv11-n was the best-performing model in terms of combined detection efficiency, model size, and computational complexity, making it a promising candidate for embedded real-time HDD. YOLOv5-s and v11-s were found to balance detection accuracy and model lightweightness, although their efficiency was only average. When comparing t/n and l/c versions, the changes in the backbone network of YOLOv9 had the greatest impact on detection accuracy, followed by the network depth_multiple and width_multiple of YOLOv5. The relative compression degrees of YOLOv5-n and YOLOv8-n were the highest, and v9-t achieved the greatest efficiency improvement in UAV HDD, followed by YOLOv10-n and v11-n. Full article

(This article belongs to the Special Issue AI and Smart Sensors for Intelligent Transportation Systems)

► Show Figures

Figure 1

16 pages, 3319 KB

Open AccessArticle

PyTSC: A Unified Platform for Multi-Agent Reinforcement Learning in Traffic Signal Control

by Rohit Bokade and Xiaoning Jin

Sensors 2025, 25(5), 1302; https://doi.org/10.3390/s25051302 - 20 Feb 2025

Cited by 2 | Viewed by 1288

Abstract

Multi-Agent Reinforcement Learning (MARL) presents a promising approach for addressing the complexity of Traffic Signal Control (TSC) in urban environments. However, existing platforms for MARL-based TSC research face challenges such as slow simulation speeds and convoluted, difficult-to-maintain codebases. To address these limitations, we [...] Read more.

Multi-Agent Reinforcement Learning (MARL) presents a promising approach for addressing the complexity of Traffic Signal Control (TSC) in urban environments. However, existing platforms for MARL-based TSC research face challenges such as slow simulation speeds and convoluted, difficult-to-maintain codebases. To address these limitations, we introduce PyTSC, a robust and flexible simulation environment that facilitates the training and evaluation of MARL algorithms for TSC. PyTSC integrates multiple simulators, such as SUMO and CityFlow, and offers a streamlined API, enabling researchers to explore a broad spectrum of MARL approaches efficiently. PyTSC accelerates experimentation and provides new opportunities for advancing intelligent traffic management systems in real-world applications. Full article

(This article belongs to the Special Issue AI and Smart Sensors for Intelligent Transportation Systems)

► Show Figures

Figure 1

25 pages, 7195 KB

Open AccessArticle

A Comprehensive Framework for Evaluating Cycling Infrastructure: Fusing Subjective Perceptions with Objective Data

by Kefei Tian, Yifan Zheng, Zhongyu Sun, Zishun Yin, Kai Zhu and Chenglong Liu

Sensors 2025, 25(4), 1168; https://doi.org/10.3390/s25041168 - 14 Feb 2025

Cited by 2 | Viewed by 1407

Abstract

As cities increasingly prioritize green and low-carbon transportation, the development of effective cycling infrastructure has become essential for alleviating traffic congestion and reducing environmental impacts. However, the service quality of bike lanes remains inadequate. To address this gap, this study proposes a multi-data-fusion [...] Read more.

As cities increasingly prioritize green and low-carbon transportation, the development of effective cycling infrastructure has become essential for alleviating traffic congestion and reducing environmental impacts. However, the service quality of bike lanes remains inadequate. To address this gap, this study proposes a multi-data-fusion framework for evaluating bike lane “cycling friendliness”, integrating subjective perceptions with objective metrics. The framework combines survey-based subjective data with digital measurements to enable rapid, large-scale assessments that align with user expectations. Tailored evaluation models are developed based on revealed preference (RP) survey analysis to account for variations among target user groups. Key factors such as road roughness, motor vehicle encroachment, cycling-friendly amenities, and roadside scenery are quantitatively assessed using vibration analysis and computer vision techniques. Validation results reveal a strong correlation between model predictions and subjective evaluations, demonstrating the framework’s reliability and effectiveness. This approach offers a scalable, data-driven tool for optimizing bike route selection and guiding infrastructure upgrades, thus advancing urban cycling transportation. Full article

(This article belongs to the Special Issue AI and Smart Sensors for Intelligent Transportation Systems)

► Show Figures

Figure 1

27 pages, 7360 KB

Open AccessArticle

Real-Time Turning Movement, Queue Length, and Traffic Density Estimation and Prediction Using Vehicle Trajectory and Stationary Sensor Data

by Amr K. Shafik and Hesham A. Rakha

Sensors 2025, 25(3), 830; https://doi.org/10.3390/s25030830 - 30 Jan 2025

Viewed by 1997

Abstract

This paper introduces a two-stage adaptive Kalman filter algorithm to estimate and predict traffic states required for real-time traffic signal control. Leveraging probe vehicle trajectory and upstream detector data, turning movement (TM) counts in the vicinity of signalized intersections are estimated in the [...] Read more.

This paper introduces a two-stage adaptive Kalman filter algorithm to estimate and predict traffic states required for real-time traffic signal control. Leveraging probe vehicle trajectory and upstream detector data, turning movement (TM) counts in the vicinity of signalized intersections are estimated in the first stage, while the upstream approach density and queue sizes are estimated in the second stage. The proposed approach is evaluated using drone-collected and simulated data from a four-legged signalized intersection in Orlando, Florida. The performance of the two-stage approach is quantified relative to the baseline estimation without a Kalman filter. The results show that the Kalman filter is effective in enhancing traffic state estimates at various market penetration levels, where the filter both improves the estimation accuracy over the baseline case and provides reliable state predictions. In the first stage, the standard deviation (SD) in TM estimates improves by up to 50% compared to the estimates provided by the sole use of probe vehicle headings. The proposed approach also provides predictions with a minimal SD of 92.8 veh/h at a 5% level of market penetration. In the second stage, the proposed queue size estimation method results in an enhancement to the queue size estimation of up to 32.8% compared to the estimates obtained from the baseline approach. In addition, the estimated traffic density is enhanced by up to 18.5%. The proposed two-stage approach demonstrates the capability of providing reliable turning movement predictions across varying levels of market penetration. This highlights the readiness of this approach for practical application in real-time traffic signal control systems. Full article

(This article belongs to the Special Issue AI and Smart Sensors for Intelligent Transportation Systems)

► Show Figures

Figure 1

21 pages, 5852 KB

Open AccessArticle

Study on the Attribute Characteristics of Road Cracks Detected by Ground-Penetrating Radar

by Shili Guo, Mingyu Yu, Zhiwei Xu, Guanghua Yue, Wencai Cai and Pengfei Tian

Sensors 2025, 25(3), 595; https://doi.org/10.3390/s25030595 - 21 Jan 2025

Cited by 2 | Viewed by 1313

Abstract

Cracks are a common form of road distress that can significantly impact pavement integrity. Accurate detection of the attribute characteristics of cracks, including the type, location (top and bottom), width, and orientation, is crucial for effective repair and treatment. This study combines numerical [...] Read more.

Cracks are a common form of road distress that can significantly impact pavement integrity. Accurate detection of the attribute characteristics of cracks, including the type, location (top and bottom), width, and orientation, is crucial for effective repair and treatment. This study combines numerical simulations with filed data to investigate how the amplitudes of ground-penetrating radar (GPR) early-time signals (ETSs) vary with changes in the crack top and width, as well as how variations in the crack bottom impact radar reflected wave amplitude. The results show that when GPR ETSs are mixed with diffracted waves from the crack top, the amplitude change percentage of the ETS at the crack top exhibits a pronounced ‘∨’-shaped dip, which provides a clearer indication of the crack top. Furthermore, a positive correlation exists between crack width and the amplitude change percentage, offering a theoretical basis for quantitatively estimating crack width. On the reflected wave originating from the interface between the semi-rigid base and the subgrade, a pronounced ‘∧’-shaped dip is observed in the trough amplitude change percentage of the reflected wave at the crack bottom. For cracks of the same width, the amplitude of the ‘∧’ vertex from reflective cracks is approximately three times greater than that from fatigue cracks. This discrepancy helps identify the crack bottom and quantitatively diagnose their types. The line connecting the vertices of the ‘∨’ and ‘∧’ shapes indicate the crack’s orientation. Accurate diagnosis of crack properties can guide precise, minimally invasive treatment methods, effectively repairing road cracks and extending the road’s service life. Full article

(This article belongs to the Special Issue AI and Smart Sensors for Intelligent Transportation Systems)

► Show Figures

Graphical abstract

Journal Menu

Journal Browser

AI and Smart Sensors for Intelligent Transportation Systems

Share This Special Issue

Special Issue Editors

Special Issue Information

Keywords

Benefits of Publishing in a Special Issue

Published Papers (12 papers)

Research

Further Information

Guidelines

MDPI Initiatives

Follow MDPI