The rapid improvement in 6G-enabled Autonomous Transport Systems (ATS) has enhanced operational efficiency in terms of communication speed, data processing, and vehicle coordination. However, it presents a critical challenge in enabling vehicles to handle unforeseen, real-time adverse conditions. Despite these advancements, the challenge of adapting to unpredictable traffic scenarios and operational anomalies persists, and there is still room for improvement in managing these situations without compromising decision-making or resource management. We propose the Distributed Intelligence Framework (DIF), which leverages Federated Learning (FL) and Digital Twins (DTs) to enhance decision-making and network resilience. FL enables collaborative learning among vehicles while ensuring sensitive data remains localized, and DTs simulate adverse traffic scenarios in real time, allowing proactive adjustments to resource allocation and traffic management. The DIF framework enables vehicles to learn from the experiences of others, allowing them to handle unique or adverse conditions that individual vehicles may not have encountered before. This collaborative approach strengthens the system’s ability to adapt to new challenges while safeguarding data integrity and ensuring operational efficiency. Experimental results show that DIF achieves a 65% reduction in convergence error within just five epochs, demonstrating significant improvements in both network resilience and decision-making, making it a critical advancement for the future of 6G-enabled ATS networks.
{"title":"Federated Learning and Digital Twin-Enabled Distributed Intelligence Framework for 6G Autonomous Transport Systems","authors":"Arikumar K. Selvaraj;Yeshwanth Govindarajan;Sahaya Beni Prathiba;A. Aashish Vinod;Vishal Pranav Amirtha Ganesan;Zhu Zhu;Thippa Reddy Gadekallu","doi":"10.1109/TITS.2025.3587448","DOIUrl":"https://doi.org/10.1109/TITS.2025.3587448","url":null,"abstract":"The rapid improvement in 6G-enabled Autonomous Transport Systems (ATS) has enhanced operational efficiency in terms of communication speed, data processing, and vehicle coordination. However, it presents a critical challenge in enabling vehicles to handle unforeseen, real-time adverse conditions. Despite these advancements, the challenge of adapting to unpredictable traffic scenarios and operational anomalies persists, and there is still room for improvement in managing these situations without compromising decision-making or resource management. We propose the Distributed Intelligence Framework (DIF), which leverages Federated Learning (FL) and Digital Twins (DTs) to enhance decision-making and network resilience. FL enables collaborative learning among vehicles while ensuring sensitive data remains localized, and DTs simulate adverse traffic scenarios in real time, allowing proactive adjustments to resource allocation and traffic management. The DIF framework enables vehicles to learn from the experiences of others, allowing them to handle unique or adverse conditions that individual vehicles may not have encountered before. This collaborative approach strengthens the system’s ability to adapt to new challenges while safeguarding data integrity and ensuring operational efficiency. Experimental results show that DIF achieves a 65% reduction in convergence error within just five epochs, demonstrating significant improvements in both network resilience and decision-making, making it a critical advancement for the future of 6G-enabled ATS networks.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 10","pages":"18214-18224"},"PeriodicalIF":8.4,"publicationDate":"2025-07-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145384610","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The assessment of road conditions is important in ensuring the safety and efficiency of transportation infrastructure. However, current methods of evaluation suffer from subjectivity, delayed response, and high costs. To address these limitations, this study proposes the development of an autonomous system that utilizes crowdsourcing RGB-D data to comprehensively and efficiently assess road conditions. The system includes an affordable vehicle-based data acquisition system, allowing for the collection of 3D pavement surface data while driving. By utilizing RGB-D sensors, the system captures both 2D color and 3D depth information of the entire lane width, with precise frame registration facilitated by a high-precision global positioning system (GPS) sensor. To evaluate the pavement conditions, the Pavement Surface Evaluation Rating (PASER) for asphalt pavement is used as a case study in this study. The establishment of an expert system for pavement condition evaluation involves the classification and quantification of pavement data using deep learning and computer vision approaches. The pavement surface data is classified into eight classes, including healthy surface, open joint, manhole, crack sealant, transverse crack, longitudinal crack, alligator cracking, and pothole. Moreover, the quantification results provide detailed information on distress and offer a more accurate understanding of pavement conditions. The system also facilitates the tracking of identified defects and repair work, providing up-to-date information on pavement deterioration and maintenance. It can be used for quality control of the pavement rehabilitation processes where the road authorities can evaluate the quality of the work that is done by the contractors.
{"title":"Affordable, Autonomous, and Comprehensive Road Condition Assessment Using RGB-D Sensors: Enhancing Pavement Condition Evaluation","authors":"Yu-Ting Huang;Mohammad Reza Jahanshahi;Nikkhil Vijaya Sankar;Fangjia Shen","doi":"10.1109/TITS.2025.3588921","DOIUrl":"https://doi.org/10.1109/TITS.2025.3588921","url":null,"abstract":"The assessment of road conditions is important in ensuring the safety and efficiency of transportation infrastructure. However, current methods of evaluation suffer from subjectivity, delayed response, and high costs. To address these limitations, this study proposes the development of an autonomous system that utilizes crowdsourcing RGB-D data to comprehensively and efficiently assess road conditions. The system includes an affordable vehicle-based data acquisition system, allowing for the collection of 3D pavement surface data while driving. By utilizing RGB-D sensors, the system captures both 2D color and 3D depth information of the entire lane width, with precise frame registration facilitated by a high-precision global positioning system (GPS) sensor. To evaluate the pavement conditions, the Pavement Surface Evaluation Rating (PASER) for asphalt pavement is used as a case study in this study. The establishment of an expert system for pavement condition evaluation involves the classification and quantification of pavement data using deep learning and computer vision approaches. The pavement surface data is classified into eight classes, including healthy surface, open joint, manhole, crack sealant, transverse crack, longitudinal crack, alligator cracking, and pothole. Moreover, the quantification results provide detailed information on distress and offer a more accurate understanding of pavement conditions. The system also facilitates the tracking of identified defects and repair work, providing up-to-date information on pavement deterioration and maintenance. It can be used for quality control of the pavement rehabilitation processes where the road authorities can evaluate the quality of the work that is done by the contractors.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 9","pages":"13094-13108"},"PeriodicalIF":8.4,"publicationDate":"2025-07-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145315451","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-07-21DOI: 10.1109/TITS.2025.3588549
Ning Wang;Yuli Hou;Chidong Qiu;Zaijin You
If underactuated dynamics can not be accommodated in path planning for an uncrewed surface vehicle (USV), sway unactuation makes the path untrackable, thereby threatening navigation resilience. In this paper, an underactuated navigation actor-critic (UNAC) deep reinforcement learning (DRL) framework is devoted to feasibly trackable path planner for an underactuated USV. By integrating a long short-term memory module into the critic network, historical state sequences are compressed into low-dimensional representations, thereby balancing optimization efficiency and complexity. To incrementally optimize pertinent path, a composite reward function covering process, collision and target approaching is created to fertilize the optimizer. Within the algorithmic flow, successive waypoints-tracking mechanism is embedded, ensuring that path-planning policy can be compatible with unactuated sway dynamics. To provide sufficiently diversified learning scenarios that can hardly experience in practice, Unity3D-based virtual-reality environments are established by replicating real-world shallow and congested situations, showcasing that the UNAC-based path planner works resiliently under unfamiliar circumstances. Compared to conventional DRL methods, the UNAC-DRL framework not only accelerates the learning process but also achieves a success rate improvement of up to 15%.
{"title":"Underactuated Navigation Actor-Critic Deep Reinforcement Learning Framework for Holistic Path Planning of Uncrewed Surface Vehicles","authors":"Ning Wang;Yuli Hou;Chidong Qiu;Zaijin You","doi":"10.1109/TITS.2025.3588549","DOIUrl":"https://doi.org/10.1109/TITS.2025.3588549","url":null,"abstract":"If underactuated dynamics can not be accommodated in path planning for an uncrewed surface vehicle (USV), sway unactuation makes the path untrackable, thereby threatening navigation resilience. In this paper, an underactuated navigation actor-critic (UNAC) deep reinforcement learning (DRL) framework is devoted to feasibly trackable path planner for an underactuated USV. By integrating a long short-term memory module into the critic network, historical state sequences are compressed into low-dimensional representations, thereby balancing optimization efficiency and complexity. To incrementally optimize pertinent path, a composite reward function covering process, collision and target approaching is created to fertilize the optimizer. Within the algorithmic flow, successive waypoints-tracking mechanism is embedded, ensuring that path-planning policy can be compatible with unactuated sway dynamics. To provide sufficiently diversified learning scenarios that can hardly experience in practice, Unity3D-based virtual-reality environments are established by replicating real-world shallow and congested situations, showcasing that the UNAC-based path planner works resiliently under unfamiliar circumstances. Compared to conventional DRL methods, the UNAC-DRL framework not only accelerates the learning process but also achieves a success rate improvement of up to 15%.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 11","pages":"21245-21256"},"PeriodicalIF":8.4,"publicationDate":"2025-07-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145486513","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-07-21DOI: 10.1109/TITS.2025.3588356
Tianyi Li;Alexander Halatsis;Raphael Stern
This paper introduces RACER, the Rational Artificial Intelligence Car-following model Enhanced by Reality, a cutting-edge deep learning car-following model, that satisfies partial derivative constraints, designed to predict Adaptive Cruise Control (ACC) driving behavior while staying theoretically feasible. Unlike conventional models, RACER effectively integrates Rational Driving Constraints (RDCs), crucial tenets of actual driving, resulting in strikingly accurate and realistic predictions. Against established models like the Optimal Velocity Relative Velocity (OVRV), a car-following Neural Network (NN), and a car-following Physics-Informed Neural Network (PINN), RACER excels across key metrics, such as acceleration, velocity, and spacing. Notably, it displays a perfect adherence to the RDCs, registering zero violations, in stark contrast to other models. This study highlights the immense value of incorporating physical constraints within AI models, especially for augmenting safety measures in transportation. It also paves the way for future research to test these models against human driving data, with the potential to guide safer and more rational driving behavior. The versatility of the proposed model, including its potential to incorporate additional derivative constraints and broader architectural applications, enhances its appeal and broadens its impact within the scientific community.
{"title":"RACER: Rational Artificial Intelligence Car-Following-Model Enhanced by Reality","authors":"Tianyi Li;Alexander Halatsis;Raphael Stern","doi":"10.1109/TITS.2025.3588356","DOIUrl":"https://doi.org/10.1109/TITS.2025.3588356","url":null,"abstract":"This paper introduces RACER, the Rational Artificial Intelligence Car-following model Enhanced by Reality, a cutting-edge deep learning car-following model, that satisfies partial derivative constraints, designed to predict Adaptive Cruise Control (ACC) driving behavior while staying theoretically feasible. Unlike conventional models, RACER effectively integrates Rational Driving Constraints (RDCs), crucial tenets of actual driving, resulting in strikingly accurate and realistic predictions. Against established models like the Optimal Velocity Relative Velocity (OVRV), a car-following Neural Network (NN), and a car-following Physics-Informed Neural Network (PINN), RACER excels across key metrics, such as acceleration, velocity, and spacing. Notably, it displays a perfect adherence to the RDCs, registering zero violations, in stark contrast to other models. This study highlights the immense value of incorporating physical constraints within AI models, especially for augmenting safety measures in transportation. It also paves the way for future research to test these models against human driving data, with the potential to guide safer and more rational driving behavior. The versatility of the proposed model, including its potential to incorporate additional derivative constraints and broader architectural applications, enhances its appeal and broadens its impact within the scientific community.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 11","pages":"21199-21214"},"PeriodicalIF":8.4,"publicationDate":"2025-07-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145486502","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
A bike-sharing system (BSS) is easily unbalanced due to the uncertainty of user demand at each bike station during the day, which appeals for an effective bike reposition solution based on the accurate prediction of user demand. However, there is a discrepancy between the bike pickup/drop-off record (satisfied demand) and the user’s first choice of origin/destination stations (i.e., true user demand) since the BSS cannot capture the unsatisfied user demand (i.e., abandoned rentals and transferred rentals/returns) that occurs at either empty stations (failed rentals) or full stations (failed returns). To efficiently rebalance the BSS, this paper focuses on accurately forecasting the true user demand of the BSS. First, we extract the spatial-temporal features of bike usage and establish a spatio-temporal model for true user demand prediction. Then, an XGBoost-based three-stage prediction approach is proposed to accurately predict the true user demand including the station clustering, the system record rectification, and the true user demand prediction. The real data from the Citi Bike in New York is applied to verify the proposed method and the experimental results demonstrate that the proposed approach outperforms the existing methods.
{"title":"An XGBoost-Based Three-Stage Prediction Approach for True User Demand of Bike-Sharing Systems Based on Spatio-Temporal Analysis","authors":"Hongfei Guo;Shuman Zhao;Yaping Ren;Jianqing Li;Suxiu Xu","doi":"10.1109/TITS.2025.3586747","DOIUrl":"https://doi.org/10.1109/TITS.2025.3586747","url":null,"abstract":"A bike-sharing system (BSS) is easily unbalanced due to the uncertainty of user demand at each bike station during the day, which appeals for an effective bike reposition solution based on the accurate prediction of user demand. However, there is a discrepancy between the bike pickup/drop-off record (satisfied demand) and the user’s first choice of origin/destination stations (i.e., true user demand) since the BSS cannot capture the unsatisfied user demand (i.e., abandoned rentals and transferred rentals/returns) that occurs at either empty stations (failed rentals) or full stations (failed returns). To efficiently rebalance the BSS, this paper focuses on accurately forecasting the true user demand of the BSS. First, we extract the spatial-temporal features of bike usage and establish a spatio-temporal model for true user demand prediction. Then, an XGBoost-based three-stage prediction approach is proposed to accurately predict the true user demand including the station clustering, the system record rectification, and the true user demand prediction. The real data from the Citi Bike in New York is applied to verify the proposed method and the experimental results demonstrate that the proposed approach outperforms the existing methods.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 9","pages":"12987-12998"},"PeriodicalIF":8.4,"publicationDate":"2025-07-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145315339","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-07-18DOI: 10.1109/TITS.2025.3587612
Daoming Wan;Jimmy Xiangji Huang
Misbehavior detection systems (MDS) play a crucial role in vehicular ad hoc networks (VANETs) to guarantee their secure operation. Most recent studies focus on applying machine learning methods to detect misbehavior messages. However, these studies are mainly proposed in the complete VANETs datasets. The MDS with incomplete messages (IMDS) is an inevitable issue that was rarely discussed in previous studies. This survey paper aims to explore the performance of missing value treatments in IMDS. It comprehensively introduces the current missing value treatments as well as the data amputation method from previous studies. Furthermore, this survey simulates the incomplete environments for VANETs datasets and conducts experiments over simulated incomplete datasets with various performance metrics. The experimental results are analyzed and discussed to indicate the best match of missing value treatments for IMDS. Finally, the potential challenges and promising future research directions are also highlighted in this paper.
{"title":"Missing Value Treatments for Machine Learning-Based Misbehavior Detection Systems: Survey, Evaluation, and Challenges","authors":"Daoming Wan;Jimmy Xiangji Huang","doi":"10.1109/TITS.2025.3587612","DOIUrl":"https://doi.org/10.1109/TITS.2025.3587612","url":null,"abstract":"Misbehavior detection systems (MDS) play a crucial role in vehicular ad hoc networks (VANETs) to guarantee their secure operation. Most recent studies focus on applying machine learning methods to detect misbehavior messages. However, these studies are mainly proposed in the complete VANETs datasets. The MDS with incomplete messages (IMDS) is an inevitable issue that was rarely discussed in previous studies. This survey paper aims to explore the performance of missing value treatments in IMDS. It comprehensively introduces the current missing value treatments as well as the data amputation method from previous studies. Furthermore, this survey simulates the incomplete environments for VANETs datasets and conducts experiments over simulated incomplete datasets with various performance metrics. The experimental results are analyzed and discussed to indicate the best match of missing value treatments for IMDS. Finally, the potential challenges and promising future research directions are also highlighted in this paper.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 9","pages":"12798-12818"},"PeriodicalIF":8.4,"publicationDate":"2025-07-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145315486","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Point cloud completion concerns the inference of the completed geometries for real-scanned point clouds that are sparse and incomplete due to occlusion, noise, and viewpoint. Previous methods usually learn a one-shot partial-to-complete mapping, which is incapable of generating fine structure details for the complex point cloud distributions. In this paper, a Hierarchical Geometry Generation point completion Network (HGG-Net) is proposed to hierarchically generate the fine-grained completed point cloud with a skeleton-to-details strategy, which consists of three fundamental modules, namely Transformer-enhanced Feature Encoder (TFE), Multi-level Geometry Representation Decoder (MGRD), and Hierarchical Dynamic Geometry Generator (HDG). Specifically, TFE first extracts geometry features of the incomplete input and obtains a coarse prediction via self-attention mechanism and edge convolution. Second, MGRD obtains the multi-level decoded geometry representations by Geometric Interactive Transformer (GIT) and Channel-Attention-based Geometry Features Fusion (CAGF), where GIT is proposed to decode the complete prompt by capturing the semantic relationship between geometry features of the incomplete and the decoded complete objects, and CAGF aims to fuse them for the high-quality representation. Third, HDG generates the complete points hierarchically from skeleton to details based on the Dynamic Graph Attention mechanism. Qualitative and quantitative experiments demonstrate that the proposed HGG-Net outperforms state-of-the-art methods on several point cloud completion datasets. Our code is available at https://github.com/haalexx/HGGNet.
{"title":"HGG-Net: Hierarchical Geometry Generation Network for Point Cloud Completion","authors":"Hao Liang;Zhaoshui He;Xu Wang;Wenqing Su;Ji Tan;Shengli Xie","doi":"10.1109/TITS.2025.3584474","DOIUrl":"https://doi.org/10.1109/TITS.2025.3584474","url":null,"abstract":"Point cloud completion concerns the inference of the completed geometries for real-scanned point clouds that are sparse and incomplete due to occlusion, noise, and viewpoint. Previous methods usually learn a one-shot partial-to-complete mapping, which is incapable of generating fine structure details for the complex point cloud distributions. In this paper, a Hierarchical Geometry Generation point completion Network (HGG-Net) is proposed to hierarchically generate the fine-grained completed point cloud with a skeleton-to-details strategy, which consists of three fundamental modules, namely Transformer-enhanced Feature Encoder (TFE), Multi-level Geometry Representation Decoder (MGRD), and Hierarchical Dynamic Geometry Generator (HDG). Specifically, TFE first extracts geometry features of the incomplete input and obtains a coarse prediction via self-attention mechanism and edge convolution. Second, MGRD obtains the multi-level decoded geometry representations by Geometric Interactive Transformer (GIT) and Channel-Attention-based Geometry Features Fusion (CAGF), where GIT is proposed to decode the complete prompt by capturing the semantic relationship between geometry features of the incomplete and the decoded complete objects, and CAGF aims to fuse them for the high-quality representation. Third, HDG generates the complete points hierarchically from skeleton to details based on the Dynamic Graph Attention mechanism. Qualitative and quantitative experiments demonstrate that the proposed HGG-Net outperforms state-of-the-art methods on several point cloud completion datasets. Our code is available at <uri>https://github.com/haalexx/HGGNet</uri>.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 9","pages":"12999-13010"},"PeriodicalIF":8.4,"publicationDate":"2025-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145315340","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-07-14DOI: 10.1109/TITS.2025.3585801
Selim Reza;Marta Campos Ferreira;J.J.M. Machado;João Manuel R. S. Tavares
Acoustic monitoring of road traffic events is an indispensable element of Intelligent Transport Systems to increase their effectiveness. It aims to detect the temporal activity of sound events in road traffic auditory scenes and classify their occurrences. Current state-of-the-art algorithms have limitations in capturing long-range dependencies between different audio features to achieve robust performance. Additionally, these models suffer from external noise and variation in audio intensities. Therefore, this study proposes a spectrogram-specific transformer model employing a multi-head attention mechanism using the scaled product attention technique based on softmax in combination with Temporal Convolutional Networks to overcome these difficulties with increased accuracy and robustness. It also proposes a unique preprocessing step and a Deep Linear Projection method to reduce the dimensions of the features before passing them to the learnable Positional Encoding layer. Rather than monophonic audio data samples, stereophonic Mel-spectrogram features are fed into the model, improving the model’s robustness to noise. State-of-the-art One-dimensional Convolutional Neural Networks and Long Short-term Memory models were used to compare the proposed model’s performance on two well-known datasets. The results demonstrated its superior performance by achieving an improvement in accuracy of 1.51 to 3.55% compared to the studied baselines.
{"title":"Road Traffic Events Monitoring Using a Multi-Head Attention Mechanism-Based Transformer and Temporal Convolutional Networks","authors":"Selim Reza;Marta Campos Ferreira;J.J.M. Machado;João Manuel R. S. Tavares","doi":"10.1109/TITS.2025.3585801","DOIUrl":"https://doi.org/10.1109/TITS.2025.3585801","url":null,"abstract":"Acoustic monitoring of road traffic events is an indispensable element of Intelligent Transport Systems to increase their effectiveness. It aims to detect the temporal activity of sound events in road traffic auditory scenes and classify their occurrences. Current state-of-the-art algorithms have limitations in capturing long-range dependencies between different audio features to achieve robust performance. Additionally, these models suffer from external noise and variation in audio intensities. Therefore, this study proposes a spectrogram-specific transformer model employing a multi-head attention mechanism using the scaled product attention technique based on softmax in combination with Temporal Convolutional Networks to overcome these difficulties with increased accuracy and robustness. It also proposes a unique preprocessing step and a Deep Linear Projection method to reduce the dimensions of the features before passing them to the learnable Positional Encoding layer. Rather than monophonic audio data samples, stereophonic Mel-spectrogram features are fed into the model, improving the model’s robustness to noise. State-of-the-art One-dimensional Convolutional Neural Networks and Long Short-term Memory models were used to compare the proposed model’s performance on two well-known datasets. The results demonstrated its superior performance by achieving an improvement in accuracy of 1.51 to 3.55% compared to the studied baselines.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 9","pages":"13011-13024"},"PeriodicalIF":8.4,"publicationDate":"2025-07-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145315456","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-07-14DOI: 10.1109/TITS.2025.3585504
Qingyuan Shen;Haobin Jiang;Aoxue Li;Marco Cecotti;Chenhui Yin;You Gong
Planning paths for Frenet-based autonomous vehicles (AVs) at roundabouts is difficult without complete and smooth reference paths. In such situations, the interpolating curve planner is often used to create segmented reference paths from simplified geometric roundabout data. While this method ensures curvature continuity within each curve segment, the continuity at the junctions of these segments is poor. Additionally, the determination of merging and diverging point positions at roundabouts has not been thoroughly explored. This paper introduces a novel approach using 5th-order Bézier curves to plan piecewise reference paths for AVs at roundabouts. The proposed method enhances endpoint curvature continuity of the Bézier curves and improves adaptability to non-standard roundabouts. A well-designed objective function is created to optimize both the geometric continuity parameters of the Bézier curves and the positions of merging and diverging points in the circulatory roadway. This function takes into account key factors, including path length and smoothness. Case studies validate the feasibility of maintaining curvature continuity at the endpoints and the method’s ability to generalize across various scenarios, proving its effectiveness for different roundabout structures. The results also confirm the method’s efficacy in generating paths from original geometric roundabout data. Lastly, the acceptable transverse deviations between real-world trajectories and reference paths demonstrate the rationality and practical applicability of this method.
{"title":"Generating G2 Continuity Reference Paths for Autonomous Vehicles at Roundabouts","authors":"Qingyuan Shen;Haobin Jiang;Aoxue Li;Marco Cecotti;Chenhui Yin;You Gong","doi":"10.1109/TITS.2025.3585504","DOIUrl":"https://doi.org/10.1109/TITS.2025.3585504","url":null,"abstract":"Planning paths for Frenet-based autonomous vehicles (AVs) at roundabouts is difficult without complete and smooth reference paths. In such situations, the interpolating curve planner is often used to create segmented reference paths from simplified geometric roundabout data. While this method ensures curvature continuity within each curve segment, the continuity at the junctions of these segments is poor. Additionally, the determination of merging and diverging point positions at roundabouts has not been thoroughly explored. This paper introduces a novel approach using 5th-order Bézier curves to plan piecewise reference paths for AVs at roundabouts. The proposed method enhances endpoint curvature continuity of the Bézier curves and improves adaptability to non-standard roundabouts. A well-designed objective function is created to optimize both the geometric continuity parameters of the Bézier curves and the positions of merging and diverging points in the circulatory roadway. This function takes into account key factors, including path length and smoothness. Case studies validate the feasibility of maintaining curvature continuity at the endpoints and the method’s ability to generalize across various scenarios, proving its effectiveness for different roundabout structures. The results also confirm the method’s efficacy in generating paths from original geometric roundabout data. Lastly, the acceptable transverse deviations between real-world trajectories and reference paths demonstrate the rationality and practical applicability of this method.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 9","pages":"13082-13093"},"PeriodicalIF":8.4,"publicationDate":"2025-07-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145315533","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In the era of big data, intelligent transportation systems are crucial for the development of smart cities, significantly impacting urban economic growth and planning. The integration of 6G networks and digital twin technology presents unprecedented opportunities to enhance urban traffic management through real-time data synchronization and high-fidelity simulations. Accurate traffic flow prediction is vital for congestion control, intelligent route planning, and effective urban traffic management. However, existing deep learning models often struggle to capture the complex spatio-temporal dependencies and dynamic spatial relationships inherent in urban traffic data, particularly in data-scarce environments. Given the spatial heterogeneity of urban data, where dense and sparse regions coexist, improving prediction accuracy in sparse areas is critical to ensuring overall forecasting performance. To address these challenges, we propose a novel framework called 6G-Enabled Digital Twin Collaborative Traffic Flow Prediction (DT-CTFP), which integrates advanced deep learning models within a 6G-supported digital twin environment. The framework leverages real-time data processing capabilities and ultra-low latency of 6G networks to capture complex traffic features and dynamic spatial dependencies. In data-rich regions, the Dynamic Graph Multi-Attention (DGMA) model is used to learn fine-grained spatio-temporal patterns, while for data-scarce regions, the Cross-Area Transfer Prediction (CATP) model utilizes meta-learning techniques to transfer knowledge from data-rich urban areas, improving prediction accuracy in areas with limited data. Experimental results demonstrate the superiority of the DT-CTFP framework, achieving up to 6% reductions in RMSE and 4% reductions in MAE across multiple datasets, highlighting its enhanced prediction accuracy and efficiency. These results emphasize the framework’s capacity to improve traffic management and vehicle-road cooperation within a digital twin smart city.
{"title":"DT-CTFP: 6G-Enabled Digital Twin Collaborative Traffic Flow Prediction","authors":"Baofu Wu;Jilin Zhang;Junfeng Yuan;Yan Zeng;Peng Zhan;Yuyu Yin;Jian Wan;Honghao Gao","doi":"10.1109/TITS.2025.3582356","DOIUrl":"https://doi.org/10.1109/TITS.2025.3582356","url":null,"abstract":"In the era of big data, intelligent transportation systems are crucial for the development of smart cities, significantly impacting urban economic growth and planning. The integration of 6G networks and digital twin technology presents unprecedented opportunities to enhance urban traffic management through real-time data synchronization and high-fidelity simulations. Accurate traffic flow prediction is vital for congestion control, intelligent route planning, and effective urban traffic management. However, existing deep learning models often struggle to capture the complex spatio-temporal dependencies and dynamic spatial relationships inherent in urban traffic data, particularly in data-scarce environments. Given the spatial heterogeneity of urban data, where dense and sparse regions coexist, improving prediction accuracy in sparse areas is critical to ensuring overall forecasting performance. To address these challenges, we propose a novel framework called 6G-Enabled Digital Twin Collaborative Traffic Flow Prediction (DT-CTFP), which integrates advanced deep learning models within a 6G-supported digital twin environment. The framework leverages real-time data processing capabilities and ultra-low latency of 6G networks to capture complex traffic features and dynamic spatial dependencies. In data-rich regions, the Dynamic Graph Multi-Attention (DGMA) model is used to learn fine-grained spatio-temporal patterns, while for data-scarce regions, the Cross-Area Transfer Prediction (CATP) model utilizes meta-learning techniques to transfer knowledge from data-rich urban areas, improving prediction accuracy in areas with limited data. Experimental results demonstrate the superiority of the DT-CTFP framework, achieving up to 6% reductions in RMSE and 4% reductions in MAE across multiple datasets, highlighting its enhanced prediction accuracy and efficiency. These results emphasize the framework’s capacity to improve traffic management and vehicle-road cooperation within a digital twin smart city.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 10","pages":"18129-18144"},"PeriodicalIF":8.4,"publicationDate":"2025-07-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145384612","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：