IET Intelligent Transport Systems最新文献

{"title":"On the Classification of Vehicle Cut-In Scenarios Severity: Empirical Evidence to Validate the Severity Classification of UN R157","authors":"Sandor Vass,&nbsp;Konstantinos Mattas,&nbsp;Riccardo Donà,&nbsp;Marton Pataki,&nbsp;Biagio Ciuffo,&nbsp;Maria Cristina Galassi,&nbsp;Zsolt Szalay,&nbsp;Viktor Tihanyi","doi":"10.1049/itr2.70157","DOIUrl":"https://doi.org/10.1049/itr2.70157","url":null,"abstract":"<p>The Fuzzy Safety Model (FSM), developed amongst all to support UN Regulation No. 157 on Automated Lane Keeping Systems (ALKS), provided a novel methodology for distinguishing between avoidable and non-avoidable cases of certain test scenarios for Automated Vehicles (AVs). ALKSs, restricted to highway environments, must avoid any reasonably foreseeable and preventable accident. However, beyond this capability, the FSM may also be an optimal tool to classify the difficulty level of the same traffic scenarios. To validate the FSM's ability to classify preventable scenarios according to their difficulty level, a test campaign was conducted focusing on the critical “cut-in” scenario, where another vehicle changes lanes in front of the ALKS, requiring it to decelerate to avoid a collision.</p><p>The study demonstrates the feasibility of the required tests and the FSM's effectiveness in categorising preventable cases by difficulty level. Results highlight the model's potential to plan, execute, and analyse cut-in scenarios beyond the scope of UN R157. This contribution supports the impartial assessment of AVs while addressing the challenge of representing diverse and challenging traffic conditions with a limited number of tests. The research results underscore the FSM's broader applicability for improving AV safety testing frameworks.</p>","PeriodicalId":50381,"journal":{"name":"IET Intelligent Transport Systems","volume":"20 1","pages":""},"PeriodicalIF":2.5,"publicationDate":"2026-03-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ietresearch.onlinelibrary.wiley.com/doi/epdf/10.1049/itr2.70157","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147568343","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Vehicle-Composition Dependent Insights Into Wide Moving Jam Propagation Using Aerial Traffic Observations","authors":"Marc Zürn,&nbsp;Hubert Rehborn,&nbsp;Lennart Querfurth,&nbsp;Robert Hoyer","doi":"10.1049/itr2.70176","DOIUrl":"https://doi.org/10.1049/itr2.70176","url":null,"abstract":"&lt;p&gt;Aerial observations with cameras or drones are a valuable technology for observing traffic over time and space. One drone flight allows the observation of approx. 600 m and 20 min of traffic, while camera observations cover longer time periods and larger freeway segments when positioned consecutively. The study uses extracted individual vehicle trajectories with a sampling rate of 0.1 s to analyse the propagation of wide moving jams. A major objective is the analysis of the downstream front propagation: the study investigates vehicle composition dependent wide moving jam propagation parameters based on heavily congested segments on the freeway A8 in Germany and the I-24 in the US. A key finding is that the downstream front velocity depends on the vehicle composition present within the wide moving jam itself, not on the overall lane-level traffic appearance. Kerner's Three-Phase traffic theory sets the theoretical framework identifying three distinct phases: (i) free flow traffic, (ii) synchronized flow traffic and (iii) wide moving jams. All vehicle trajectories are processed with an adapted traffic state transition model with parameters refined with a sensitivity analysis. On the A8 in Germany, the results reveal the mean downstream front velocity &lt;span&gt;&lt;/span&gt;&lt;math&gt;\u0000 &lt;semantics&gt;\u0000 &lt;msub&gt;\u0000 &lt;mi&gt;v&lt;/mi&gt;\u0000 &lt;mi&gt;g&lt;/mi&gt;\u0000 &lt;/msub&gt;\u0000 &lt;annotation&gt;$v_{text{g}}$&lt;/annotation&gt;\u0000 &lt;/semantics&gt;&lt;/math&gt; of a wide moving jam for passenger cars propagates at approximately &lt;span&gt;&lt;/span&gt;&lt;math&gt;\u0000 &lt;semantics&gt;\u0000 &lt;mrow&gt;\u0000 &lt;mo&gt;−&lt;/mo&gt;\u0000 &lt;mn&gt;13&lt;/mn&gt;\u0000 &lt;/mrow&gt;\u0000 &lt;annotation&gt;$-13$&lt;/annotation&gt;\u0000 &lt;/semantics&gt;&lt;/math&gt; to &lt;span&gt;&lt;/span&gt;&lt;math&gt;\u0000 &lt;semantics&gt;\u0000 &lt;mrow&gt;\u0000 &lt;mo&gt;−&lt;/mo&gt;\u0000 &lt;mn&gt;16&lt;/mn&gt;\u0000 &lt;mspace&gt;&lt;/mspace&gt;\u0000 &lt;mtext&gt;km/h&lt;/mtext&gt;\u0000 &lt;/mrow&gt;\u0000 &lt;annotation&gt;$-16nobreakspace text{km/h}$&lt;/annotation&gt;\u0000 &lt;/semantics&gt;&lt;/math&gt; while truck-only lanes show a tendency toward considerably faster downstream velocities of approximately &lt;span&gt;&lt;/span&gt;&lt;math&gt;\u0000 &lt;semantics&gt;\u0000 &lt;mrow&gt;\u0000 &lt;mo&gt;−&lt;/mo&gt;\u0000 &lt;mn&gt;21&lt;/mn&gt;\u0000 &lt;/mrow&gt;\u0000 &lt;annotation&gt;$-21$&lt;/annotation&gt;\u0000 &lt;/semantics&gt;&lt;/math&gt; to &lt;span&gt;&lt;/span&gt;&lt;math&gt;\u0000 &lt;semantics&gt;\u0000 &lt;mrow&gt;\u0000 &lt;mo&gt;−&lt;/mo&gt;\u0000 &lt;mn&gt;24&lt;/mn&gt;\u0000 &lt;mspace&gt;&lt;/mspace&gt;\u0000 &lt;mtext&gt;km/h&lt;/mtext&gt;\u0000 &lt;/mrow&gt;\u0000 &lt;annotation&gt;$-24nobreakspace text{km/h}$&lt;/annotation&gt;\u0000 &lt;/semantics&gt;&lt;/math&gt;. Wide moving jams with mixed traffic appearance show intermediate downstream front velocities. Despit","PeriodicalId":50381,"journal":{"name":"IET Intelligent Transport Systems","volume":"20 1","pages":""},"PeriodicalIF":2.5,"publicationDate":"2026-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ietresearch.onlinelibrary.wiley.com/doi/epdf/10.1049/itr2.70176","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147665887","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Intention-Aware Electric Vehicle Routing: A Markovian Queuing Model Approach to Reduce Wait Times at Charging Stations","authors":"Julian Widmann,&nbsp;Oliver Sawodny","doi":"10.1049/itr2.70180","DOIUrl":"https://doi.org/10.1049/itr2.70180","url":null,"abstract":"<p>As battery electric vehicles become increasingly common, the availability of high-power charging stations for long-distance travel has become critical. However, the expansion of charging infrastructure has not kept pace with rising demand, resulting in queues and wait times during peak periods. This paper presents a comprehensive analysis of the intention-aware fixed-route electric vehicle charging problem. We show that selfish routing algorithms, which optimize decisions at the individual-vehicle level, often lead to suboptimal system-wide performance. To address this, we propose an intention-aware queuing system based on Markov models to improve charging station assignment. The system relies on vehicles submitting their planned charging intentions and, in return, provides wait time predictions that are integrated into a dynamic programming-based charging strategy optimization framework. We evaluate the effectiveness of the approach through a large-scale simulation using real-world electric vehicle mobility data from a major German automotive manufacturer. The simulation includes up to 7500 simultaneously operating vehicles across Germany and is implemented in MATSim. The results show reductions in waiting times of up to 96%, demonstrating substantial improvements in overall system performance and efficiency. These findings highlight the importance of collaborative and coordinated strategies for optimizing electric vehicle charging and routing under high demand conditions.</p>","PeriodicalId":50381,"journal":{"name":"IET Intelligent Transport Systems","volume":"20 1","pages":""},"PeriodicalIF":2.5,"publicationDate":"2026-03-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ietresearch.onlinelibrary.wiley.com/doi/epdf/10.1049/itr2.70180","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147665890","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Investigation of Blockchain for Cybersecurity in Maritime Autonomous Surface Ships","authors":"Ian Hamilton,&nbsp;Dante Ben Matellini,&nbsp;Chia-Hsun Chang","doi":"10.1049/itr2.70184","DOIUrl":"https://doi.org/10.1049/itr2.70184","url":null,"abstract":"<p>As Maritime Autonomous Surface Ships (MASS) become increasingly prevalent, cybersecurity emerges as a critical challenge due to the reliance on interconnected digital systems. Blockchain technology offers promising solutions to enhance data security, transparency and operational resilience. However, the high energy consumption of blockchain creates a dilemma for the maritime sector to meet low-carbon and sustainability targets. Identifying strategies balancing energy consumption with the security benefits brought by blockchain is therefore essential. This research aims to examine the integration of blockchain to improve cybersecurity, operational efficiency and data integrity in MASS. The study contributes to an integration of a literature review, strengths, weaknesses, opportunities, threats (SWOT), threats, opportunities, weaknesses, strengths (TOWS) analysis with a practitioner survey, providing strategic insights into blockchain to strengthen cybersecurity frameworks for MASS. Findings of SWOT-TOWS highlight blockchain's potential to improve cybersecurity and automate functions through smart contracts, while identifying significant barriers such as scalability, energy consumption, etc. Survey results show strong industry support for blockchain's role in enhancing cybersecurity, particularly in communication and automation systems. Respondents emphasised the importance of protocols and transaction speed. The study concludes with strategic recommendations and policy considerations to guide effective blockchain integration in MASS.</p>","PeriodicalId":50381,"journal":{"name":"IET Intelligent Transport Systems","volume":"20 1","pages":""},"PeriodicalIF":2.5,"publicationDate":"2026-03-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ietresearch.onlinelibrary.wiley.com/doi/epdf/10.1049/itr2.70184","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147665891","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Attention-Guided Lightweight CNN-Transformer Fusion for Real-Time Traffic Sign Recognition in Adverse Environments: HACTNet","authors":"Mandeep Singh Devgan,&nbsp;Gurvinder Singh,&nbsp;Purushottam Sharma,&nbsp;Tajinder Kumar,&nbsp;Xiaochun Cheng,&nbsp;Deepak Ahlawat","doi":"10.1049/itr2.70167","DOIUrl":"https://doi.org/10.1049/itr2.70167","url":null,"abstract":"<p>Autonomous driving is also impossible without traffic sign recognition (TSR; also known as traffic sign-on-road), which limits its reliability to domain changes, unfavourable weather, obstruction and hardware capacity. This paper proposes HACTNet, a low-complexity CNN-Transformer hybrid model that pushes the state-of-art in TSR by making a noteworthy set of contributions including (i) efficient convaps to model parts of the image, (ii) transformer encoder to capture the global context and (iii) an attention-based fusion block to dynamically combine the two complementary sets of features. This synergy facilitates strong recognition in presence of blur and occlusion and in varying illumination. In addition to accuracy, HACTNet achieves high robustness (52.8%) against strong PGD adversarial attacks (8/255), but is still efficient (7.9 M parameters and 22.1 FPS) on the NVIDIA Jetson Nano. Moreover, the comparative analysis between the hybrid models (EATFormer, local-ViT) and HACTNet proves that HACTNet has a better accuracy-efficiency ratio. The extraordinary capability to counteract adverse weather conditions, fog, night, rain, snow etc., which is proven by the extensive testing of the real-world ACDC adverse conditions data set, supports the viability of the proposed solutions in the real world. It is plug and play modularity with on-going learning via elastic weight consolidation (3.3% less forgetting) and unsupervised domain adaptation via MMD loss (5.3% better on TT100K with no labels). Moreover, INT8 quantization with quantization-aware training (QAT) incurs little accuracy loss (less than 0.5 percent) and much lower energy (0.27 J/sample) usage, which forms an edge deployment preparedness. Additionally, when adjusting to new traffic signs over time, the model shows compatibility with continuous learning, achieving a low forgetting rate (3.3%), highlighting its practical viability for long-term autonomous deployment. Overall, HACTNet produces a versatile and expandable solution for next-generation intelligent transportation systems by striking a balance between accuracy, robustness and efficiency.</p>","PeriodicalId":50381,"journal":{"name":"IET Intelligent Transport Systems","volume":"20 1","pages":""},"PeriodicalIF":2.5,"publicationDate":"2026-02-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ietresearch.onlinelibrary.wiley.com/doi/epdf/10.1049/itr2.70167","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147666178","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Graph Neural Network-Enabled Robust AIS/Radar Data Association for Intelligent Monitoring of Ship Safety","authors":"Chenjie Zhao,&nbsp;Gang Liu,&nbsp;Ryan Wen Liu","doi":"10.1049/itr2.70166","DOIUrl":"https://doi.org/10.1049/itr2.70166","url":null,"abstract":"<p>In intelligent waterborne transportation systems, wide-range perception devices such as AIS receivers and navigation radars are commonly deployed to enhance the monitoring capability of ship traffic. The AIS data provides ship identity and dynamic navigational information. The radar complementarily produces higher frequency position and movement data, but it has detection blind spots. To further improve ship intelligent surveillance, it becomes necessary to associate the AIS and radar data to simultaneously capture identity, high-frequency and dynamic information for the ships of interest. However, AIS and radar data suffer from time asynchrony, spatial differences, and environmental noise. The conventional track-based AIS/radar association methods heavily rely on precise tracking results and stable trajectories, and also suffer from high computational complexity. We propose a graph neural network (GNN)-enabled robust AIS/radar data association method. The association task is formulated as a graph matching problem between imbalanced bipartite graphs. We then construct a paired graph dataset from AIS and radar data. A specialized GNN architecture is developed to extract and aggregate spatial distribution features of ship targets and their neighboring nodes. Contrastive learning techniques are employed to formulate the loss function for feature extraction networks. Subsequently, cosine similarity metrics between extracted ship distribution features are computed to construct a cost matrix for target association. The final linear assignment problem is resolved using the Hungarian algorithm, enabling precise AIS/radar data association. This methodology demonstrates significant improvements in both computational efficiency and matching accuracy.</p>","PeriodicalId":50381,"journal":{"name":"IET Intelligent Transport Systems","volume":"20 1","pages":""},"PeriodicalIF":2.5,"publicationDate":"2026-02-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ietresearch.onlinelibrary.wiley.com/doi/epdf/10.1049/itr2.70166","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147666143","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"SEDM: A Safety-Enhanced Decision-Making Framework for Autonomous Driving by Integrating Large Language Models and XGBoost","authors":"Jun Li,&nbsp;Baozhu Chen,&nbsp;Kai Xu,&nbsp;Xiaohan Yang,&nbsp;Mengting Sun,&nbsp;Guojun Li,&nbsp;Haojie Du","doi":"10.1049/itr2.70178","DOIUrl":"https://doi.org/10.1049/itr2.70178","url":null,"abstract":"<p>Large language models (LLMs) are promising for autonomous driving decision-making, but existing methods mostly rely on cloud-side deployment, causing high decision latency, privacy concerns and a lack of explicit safety verification for generated actions. To address these challenges, we propose SEDM (safety-enhanced decision-making framework) for highway driving scenarios. SEDM comprises an environment encoding module, an edge-side LLM-based decision-making module enhanced through chain-of-thought prompting and low-rank adaptation (LoRA) fine-tuning, and an XGBoost-based safety shield module that filters unsafe actions generated by the LLM. Experiments show that SEDM achieves driving success rates of 95%, 82% and 55% under simple, normal and dense traffic conditions, respectively—substantially outperforming such as deep Q-network and proximal policy optimization. Moreover, it yields a 17-percentage-point improvement in success rate over an ablated variant without the safety shield module. Furthermore, decision latency is reduced from 7.80 s (cloud-side LLM) to 1.01 s.</p>","PeriodicalId":50381,"journal":{"name":"IET Intelligent Transport Systems","volume":"20 1","pages":""},"PeriodicalIF":2.5,"publicationDate":"2026-02-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ietresearch.onlinelibrary.wiley.com/doi/epdf/10.1049/itr2.70178","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147666144","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Assessing Driver Distraction From Unmanned Aerial Vehicles and Under-Bridge Inspection Trucks: A Driving Simulator Study Using Eye Tracker and EEG","authors":"Zainab Afzali Kusha,&nbsp;Sahar Ghanipoor Machiani,&nbsp;Reza Akhavian","doi":"10.1049/itr2.70175","DOIUrl":"https://doi.org/10.1049/itr2.70175","url":null,"abstract":"<p>Unmanned aerial vehicles (UAVs) are becoming increasingly prevalent in transportation infrastructure projects due to their operational versatility and ability to minimise traffic disruptions. Consequently, federal and state agencies have shown a marked increase in interest in replacing or augmenting traditional bridge inspection methods with safer and more cost-effective approaches, such as UAV deployment. However, the potential distraction effect of UAV operations on the travelling public, particularly compared with established methods such as under-bridge inspection trucks (UBITs), has not yet been thoroughly studied. This study investigates and compares driver distraction caused by roadside bridge inspection operations using UAVs and UBITs. Through a series of controlled driving simulator experiments, driver distraction is quantified by the data collected from synchronised eye-tracking and electroencephalography sensors worn by experiment participants. The results reveal distinct distraction profiles for the two bridge inspection methods. UBIT operations consistently induced greater visual and cognitive distraction across traffic conditions, with drivers directing frequent, prolonged glances toward the prominent equipment. In contrast, UAV-related distraction was context-dependent and primarily manifested as increased visual demand in congested, low-speed traffic. Compared with baseline methods such as eye-tracking alone and other mainstream distraction assessment techniques, the proposed multimodal framework provides greater sensitivity by detecting subtle cognitive effects alongside observable visual behaviour. These findings demonstrate that UAVs generally impose lower and more situational distractions than traditional UBITs and provide evidence-based support for transportation agencies seeking to prioritise UAV deployment for routine roadway inspections while maintaining roadway safety.</p>","PeriodicalId":50381,"journal":{"name":"IET Intelligent Transport Systems","volume":"20 1","pages":""},"PeriodicalIF":2.5,"publicationDate":"2026-02-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ietresearch.onlinelibrary.wiley.com/doi/epdf/10.1049/itr2.70175","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147666111","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Vehicle Re-Identification With Diverse and Aligned Fine-Grained Features","authors":"Jian Wan,&nbsp;Yue Zeng,&nbsp;Ming Gao,&nbsp;Changshi Xiao,&nbsp;Yamin Huang,&nbsp;Huayu Liu","doi":"10.1049/itr2.70173","DOIUrl":"https://doi.org/10.1049/itr2.70173","url":null,"abstract":"<p>Vehicle re-identification (ReID), as a key task in intelligent transportation systems, has a significant impact on tracking target vehicles and developing smart cities. Vehicle ReID captures fine-grained information by decoupling local features from vehicle images. However, differences in camera viewpoints and poses can lead to substantial misalignment issues. Most existing methods align features using predefined external cues, which is inefficient and requires additional manual annotations. In this paper, we propose feature decoupled re-identification (FDReID). This model uses ResNet-50 as the backbone network to extract global features. It also includes a feature decoupling module based on the transformer structure that uses the class attention mechanism to extract fine-grained vehicle features as a supplement to the global features, and aligns the fine-grained features based on unsupervised clustering. Different from previous methods which decompose vehicle feature into structured features with the use of extra annotation or partition the feature map into several stripes or grids coarsely, we aim to make model learn to capture the discriminative fine-grained information from the flexible decomposed features. Compared with existing methods, this model can complete model training in an end-to-end manner without introducing additional annotations and additional models. In terms of experiments, compared with the StrongBaseline benchmark model, the mean average precision value of this model on the VeRi-776 dataset has increased by 7.6%, and the Rank1 and Rank5 indicators have increased by 2.6% and 1.3% respectively.</p>","PeriodicalId":50381,"journal":{"name":"IET Intelligent Transport Systems","volume":"20 1","pages":""},"PeriodicalIF":2.5,"publicationDate":"2026-02-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ietresearch.onlinelibrary.wiley.com/doi/epdf/10.1049/itr2.70173","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147666110","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Enhancing Rail Transit Safety: Edge Computing-Based Driver State Monitoring System","authors":"Yiqing Liu,&nbsp;Gaoyun Cheng,&nbsp;Binbin Yuan,&nbsp;Hongyu Sun,&nbsp;Edward Stewart,&nbsp;Zhe Fu,&nbsp;Zongru Ma,&nbsp;Ning Zhao","doi":"10.1049/itr2.70174","DOIUrl":"https://doi.org/10.1049/itr2.70174","url":null,"abstract":"<p>Ensuring the safety and alertness of railway drivers is essential, yet many existing monitoring systems still experience noticeable delays, limited real-time performance, and inefficient use of communication resources. To address these challenges, we present a rail transit driver state monitoring system that integrates edge computing with cloud-based management. The system employs multiple onboard cameras, including a cabin surveillance camera and an AI assisted camera that tracks the driver's movements, to provide a broad view of driver and cabin activity. A Jetson Orin edge unit performs real-time inference using deep learning models for driver authentication, facial-state analysis, behaviour recognition, and cabin personnel monitoring. To reduce bandwidth consumption, the system transmits compact feature representations during normal operation and uploads raw video to the cloud only when abnormal events are detected, enabling network-wide supervision via a central platform. Experiments demonstrate that the proposed edge-first pipeline achieves an average end-to-end decision latency of 173.73 ms on the edge device, while maintaining strong detection performance on the test set, reaching 96.82% accuracy for blink detection, 98.33% accuracy for yawn detection, and 97.22% accuracy for driver behaviour monitoring.</p>","PeriodicalId":50381,"journal":{"name":"IET Intelligent Transport Systems","volume":"20 1","pages":""},"PeriodicalIF":2.5,"publicationDate":"2026-02-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ietresearch.onlinelibrary.wiley.com/doi/epdf/10.1049/itr2.70174","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147666119","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}