IEEE Transactions on Intelligent Transportation Systems最新文献

{"title":"A DRL-Based Lane-Changing Decision Model for Connected and Automated Trucks to Reduce Congestion in Freeway Merging Areas","authors":"Xiaoyu He;Xumei Chen;Peikun Li;Jianchang Huang;Anran Liu;Lei Yu","doi":"10.1109/TITS.2026.3722503","DOIUrl":"https://doi.org/10.1109/TITS.2026.3722503","url":null,"abstract":"Trucks, which primarily travel in the rightmost lane on freeways, are in direct conflict with on-ramp vehicles and constitute a key factor contributing to congestion in merging areas. To address this challenge, this study proposes an adaptive information-fusion and stability-driven deep reinforcement learning (ASR) model for connected and automated trucks (CATs) lane-changing decision-making. Specifically, a graph convolutional network (GCN) is employed to encode topological relationships and interaction strengths into the state space, capturing the dynamic interactions between CATs and other vehicles in the connected traffic environment. In addition, to improve traffic system stability, a reward function based on the Lyapunov stability principle is designed to suppress fluctuations in traffic flow speed and acceleration, thereby mitigating merging conflicts. Simulation experiments conducted in Simulation of Urban MObility (SUMO) at varying traffic volumes demonstrate that the proposed model outperforms all benchmark models in alleviating merging area congestion, improving traffic stability, and reducing total travel time, total delay time, fuel consumption, and NOx emissions on the studied freeway segment. Moreover, the effectiveness of the ASR model is further validated under different CAT penetration rates, different speed limits, and unseen traffic volumes. The proposed model is also shown to improve truck lane change safety based on time-to-collision (TTC) comparisons. The findings of this study may provide useful guidance for traffic operations in freeway merging areas.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"27 10","pages":"11997-12012"},"PeriodicalIF":9.1,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097428","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}

{"title":"ReSO: A Real-Time Scalable Offloading Approach for Dynamic Vehicle-to-Everything Systems","authors":"Peng Peng;Weiwei Lin;Wentai Wu;Fang Shi;Mian Guo;Keqin Li","doi":"10.1109/TITS.2026.3708270","DOIUrl":"https://doi.org/10.1109/TITS.2026.3708270","url":null,"abstract":"Smart vehicles increasingly rely on a growing range of nearby devices, such as parked cars and roadside units, as service nodes (SNs) for offloading latency-sensitive tasks in Vehicle-to-Everything (V2X) networks. Deep reinforcement learning (DRL) has emerged as a powerful scheduler for this setting, yet dimensional mismatch at runtime renders a pre-trained agent incompatible whenever system scale or task count changes, and retraining from scratch is prohibitively costly and often impractical. To resolve this issue, we present ReSO, a DRL-based Real-time Scalable Offloading approach built on two targeted mechanisms. To eliminate task-count-induced dimension variation, we adopt sequential single-task dispatching in place of batch scheduling, casting the offloading problem as an event-driven Semi-Markov Decision Process (Semi-MDP). A homotopy reward then corrects the resulting temporal credit mis-assignment. For scalability across varying system scales, we leverage a heterogeneous graph attention encoder to produce scale-invariant embeddings, and a pointer output module to generate actions that dynamically adapt to the number of reachable SNs. Together, these mechanisms make ReSO transferable across dynamic V2X systems. Experiments confirm that, after zero-shot transfer to systems of varying scale, ReSO consistently reduces system cost by up to 35% over all baselines, even those afforded full retraining.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"27 10","pages":"12531-12543"},"PeriodicalIF":9.1,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097601","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}

{"title":"CMMS-Net: Data-Efficient Semi-Supervised Crack Segmentation With Multi-Scale Dataset Memory","authors":"Mingkun Zhu;Xiaoxia Han;Jinde Wu;Wenxin Chai;Haonan Zhu","doi":"10.1109/TITS.2026.3703727","DOIUrl":"https://doi.org/10.1109/TITS.2026.3703727","url":null,"abstract":"Automated pavement crack detection is vital for infrastructure maintenance. We propose CMMS-Net, a data-efficient semi-supervised framework that achieves engineering-grade segmentation accuracy with only 50% pixel-level annotations. Core to our approach is the Multi-Scale Dataset Memory Unit (MSCMU), which leverages dataset-level crack prototypes to preserve topological continuity and suppress background noise under limited supervision. Validated across 12 public datasets, the proposed algorithm has achieved state-of-the-art performance and even outperforms SOTA algorithms on some datasets. We demonstrate that 50% annotations represent a practical optimum for safety-critical inspection. This work establishes a reproducible benchmark for resource-constrained infrastructure monitoring in intelligent transportation systems.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"27 10","pages":"13181-13197"},"PeriodicalIF":9.1,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097741","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}

{"title":"SA-TSDTP: A State-Aware Two-Stage Framework for Logistics Drone Trajectory Prediction","authors":"Chuanqi Ma;Lingshu Zhong;Huasa Zhu;Mingyang Pei;Ming Cai","doi":"10.1109/TITS.2026.3698652","DOIUrl":"https://doi.org/10.1109/TITS.2026.3698652","url":null,"abstract":"The high flexibility and uncertainty of the flight trajectory are important factors affecting the operational safety of logistics drones. Reliable flight trajectory prediction is the core technology to solve this problem. In order to solve the multivariate time series prediction task of short-time flight trajectory prediction, this work proposes a state-aware two-stage prediction model. In the first stage, logistics drone trajectories are segmented into three flight states based on geometric characteristics and altitude change rates. In the second stage, a multi-expert fusion module incorporating flight state information is introduced to capture the underlying motion patterns of drones and enhance prediction accuracy. The proposed prediction model is evaluated using trajectory data collected from multiple real logistics drones in Shenzhen, China. Experimental results demonstrate that the proposed approach exhibits strong potential and consistently outperforms baseline methods across multiple datasets and evaluation metrics. These findings provide valuable technical support for the future development of safe urban air traffic operations and smart cities.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"27 10","pages":"13141-13154"},"PeriodicalIF":9.1,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097737","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}

{"title":"Coop-WD: Cooperative Perception With Weighting and Denoising for Robust V2V Communication","authors":"Chenguang Liu;Jianjun Chen;Yunfei Chen;Yubei He;Zhuangkun Wei;Hongjian Sun;Haiyan Lu;Qi Hao","doi":"10.1109/TITS.2026.3703542","DOIUrl":"https://doi.org/10.1109/TITS.2026.3703542","url":null,"abstract":"Cooperative perception, leveraging shared information from multiple vehicles via vehicle-to-vehicle (V2V) communication, plays a vital role in autonomous driving to alleviate the limitation of single-vehicle perception. Existing works have explored the effects of V2V communication impairments on perception precision, but they lack generalization to different levels of impairments. In this work, we propose a joint weighting and denoising framework, <italic>Coop-WD</i>, to enhance cooperative perception subject to V2V channel impairments. In this framework, the self-supervised contrastive model and the conditional diffusion probabilistic model are adopted hierarchically for vehicle-level and pixel-level feature enhancement. An efficient variant, <italic>Coop-WD-eco</i>, is also proposed to selectively deactivate denoising to reduce processing overhead. Rician fading, non-stationarity, and time-varying distortion are considered. Simulation results demonstrate that the proposed <italic>Coop-WD</i> outperforms conventional benchmarks in all cases considered. Moreover, <italic>Coop-WD</i> is validated as a plug-in module with different fusion schemes, demonstrating its generalization capability for communication-aware cooperative perception enhancement. Qualitative analysis with visual examples further proves the superiority of our proposed method. The proposed <italic>Coop-WD-eco</i> achieves up to 50% reduction in computational cost under severe distortion while maintaining comparable accuracy.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"27 10","pages":"12472-12485"},"PeriodicalIF":9.1,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097692","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}

{"title":"Contingency Planning for Autonomous Vehicles at Uncontrolled Intersections Based on Level-k Game Model Predictions","authors":"Xinpeng Chen;Binjie Mou;Zeyu Yang;Yougang Bian;Manjiang Hu;Zhijun Chen","doi":"10.1109/TITS.2026.3704219","DOIUrl":"https://doi.org/10.1109/TITS.2026.3704219","url":null,"abstract":"Uncontrolled intersections without traffic signals or lane markings (e.g., mining areas and rural roads) present critical challenges for autonomous vehicles, primarily due to complex interactive dynamics and unpredictable turning intentions. To address this, this paper proposes an integrated contingency planning framework based on level-<inline-formula> <tex-math>$k$ </tex-math></inline-formula> game model predictions. Our approach first incorporates the influence of the ego vehicle’s decisions into the prediction loop. Laterally, we represent the long-term turning intention of other vehicles via a curve-fitting method; longitudinally, we model the interactive decision-making process between the ego and surrounding vehicles using a level-<inline-formula> <tex-math>$k$ </tex-math></inline-formula> game model, thereby generating prediction and decision results that account for interaction dynamics. We then introduce a two-stage probabilistic update scheme, which separately refines the confidence of lateral and longitudinal predictions using a Gaussian Mixture Model (GMM). Finally, a motion-sampling-based contingency planner is designed to handle multimodal prediction outcomes, yielding a branched trajectory tree that is further optimized for smoothness and safety. Simulations demonstrate that our method reduces intersection traversal time by 24.6% compared to the baseline. Furthermore, it ensures collision-free maneuvers and significantly lowers maximum deceleration and jerk. The framework thus enables safer and more efficient navigation for pairwise interactions in infrastructure-less environments.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"27 10","pages":"12806-12818"},"PeriodicalIF":9.1,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097705","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}

{"title":"Robust Lane Change Decision for Autonomous Vehicles: A Causal-Invariant Adversarial Reinforcement Learning Approach","authors":"Yongjie Wang;Kaiyue Wang;Yuchen Niu;Zixuan Zhang;Qiong Li;Xinhua Mao","doi":"10.1109/TITS.2026.3698480","DOIUrl":"https://doi.org/10.1109/TITS.2026.3698480","url":null,"abstract":"Despite the rapid development of autonomous driving technology, the lane-changing polices of autonomous vehicles still face two major challenges: perceptual uncertainty and vulnerability to adversarial attacks. Existing approaches often rely on empirical regularization or adversarial training to handle input perturbations, but they lack a systematic modeling of the causal relationship between perturbation mechanisms and policy deviations, making it difficult to fundamentally improve robustness. To address these issues, this paper proposes a novel causal-invariant adversarial reinforcement learning (CIARL) approach to enhance the robustness and safety of autonomous vehicles under optimal observational perturbations. The approach primarily comprises two core modules: the attacker employs a closed-box perturbation generator based on Bayesian optimization to construct optimal adversarial perturbations in the feature representation mapping space; the defender incorporates a causal-invariant representation learning mechanism to mitigate attacks by reducing the influence of non-causal features. This mechanism constrains the impact of adversarial perturbations on policy and cost within a controllable range while maximizing rewards, thereby enabling robust policy learning. Finally, the proposed CIARL approach is evaluated in the SUMO simulator across complex highway and on-ramp scenarios. Experimental results demonstrate that our method significantly outperforms several state-of-the-art baselines in both lane-change robustness and collision safety, verifying the core role of the causal-invariant mechanism in adversarial robust policy modeling.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"27 10","pages":"12835-12848"},"PeriodicalIF":9.1,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097707","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}

{"title":"Autonomous Motion Decision-Making Considering Dual Uncertainties in Surrounding Vehicle Cut-in Intention-Action","authors":"Mingchun Cao;Chunyan Wang;Wanzhong Zhao;Ziyu Zhang","doi":"10.1109/TITS.2026.3723895","DOIUrl":"https://doi.org/10.1109/TITS.2026.3723895","url":null,"abstract":"Reliable motion decision-making is critically associated with safety, efficiency, and comfort of autonomous vehicles. However, uncertainties in the cut-in behavior of vehicles in adjacent lanes make it difficult for autonomous vehicles to accurately assess and make decisions, thereby increasing the collision risks. To address this issue, a motion decision-making framework that accounts for dual uncertainties in surrounding vehicle cut-in intention and action is proposed, aiming to enhance driving safety and adaptability in dynamic traffic environments. Specifically, a hierarchical specific intention attention network is designed to extract vehicle motion correlations to predict cut-in probabilities and map driving intention uncertainty. Then, an uncertainty-driven Gaussian Markov decision-making model is proposed, where Gaussian process regression is used to further model vehicle cut-in action uncertainty based on uncertain intention. Additionally, a deep reinforcement learning algorithm for uncertainty exploration is proposed to learn the optimal driving strategy by embedding uncertainty information into the decision-making and exploration process through a multi-criteria information-gain reward mechanism. Simulation results demonstrate that our proposed method enables effective autonomous decision-making in uncertain cut-in scenarios, and has certain advantages in terms of overall safety, efficiency and comfort compared with existing algorithms.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"27 10","pages":"12158-12172"},"PeriodicalIF":9.1,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097830","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}

{"title":"Pride and Prejudice: Toward an Information-Theoretic Framework for Mutually Communicative Driver Behavior Modeling","authors":"Tingjun Li;Nan Xu;Shuo Feng;Hassan Askari;Bruno Henrique Groenner Barbosa;Konghui Guo","doi":"10.1109/TITS.2026.3703952","DOIUrl":"https://doi.org/10.1109/TITS.2026.3703952","url":null,"abstract":"Mixed autonomy driving becomes unsafe and inefficient when autonomous vehicles (AVs) and human-driven vehicles (HVs) misread each other’s intentions. We study this problem as implicit mutual communication in lane changes. The proposed framework models how the ego vehicle both expresses its intent and probes the other driver’s preference under epistemic uncertainty. It combines a level-k Bayesian persuasion game with virtual features for proactive signaling, information-theoretic rewards for mutual communication, and adaptive weights of communication affordances. We further introduce the Pride-Inquiry (P-I) and Pride-Prejudice (P-P) planes to analyze communication intensity and tendency. The model is calibrated with a Communication-Based Multi-Agent Inverse Reinforcement Learning algorithm (C-MIRL) on the naturalistic NGSIM dataset. Compared with the non-communicative baseline, the proposed model reduces the prediction error of mandatory lane changes by up to 20% while maintaining strong generalization. Driver-In-the-Loop questionnaire scores are positively correlated with the calibrated communication variables, supporting the subjective validity of the model. The learned rewards further show that inquiry and listening affordances contribute more than pride and expression alone, and that inquiry preference varies more strongly across drivers. These results support explicit modeling of mutual communication and epistemic uncertainty in interactive driving.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"27 10","pages":"12996-13010"},"PeriodicalIF":9.1,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097894","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}

{"title":"Attention in Motion: Secure Platooning via Transformer-Based Misbehavior Detection","authors":"Konstantinos Kalogiannis;Ahmed Mohamed Hussain;Hexu Li;Panos Papadimitratos","doi":"10.1109/TITS.2026.3711275","DOIUrl":"https://doi.org/10.1109/TITS.2026.3711275","url":null,"abstract":"Vehicular platooning promises transformative improvements in transportation efficiency and safety through the coordination of multi-vehicle formations enabled by Vehicle-to-Everything (V2X) communication. However, the distributed nature of platoon coordination creates security vulnerabilities, allowing authenticated vehicles to inject falsified kinematic data, compromise operational stability, and pose a threat to passenger safety. Traditional misbehaviour detection approaches, which rely on plausibility checks and statistical methods, suffer from high False Positive (FP) rates and cannot capture the complex temporal dependencies inherent in multi-vehicle coordination dynamics. We present Attention In Motion (<sc>AIMformer</small>), a transformer-based framework specifically tailored for real-time misbehaviour detection in vehicular platoons with edge deployment capabilities. <sc>AIMformer</small> leverages multi-head self-attention mechanisms to capture intra-vehicle temporal dynamics, with a spatio-temporal variant that further models inter-vehicle spatial correlations. It incorporates global positional encoding with vehicle-specific temporal offsets to handle join/exit maneuvers. We propose a Precision-Focused Binary Cross-Entropy (PFBCE) loss function that penalizes FPs to meet the requirements of safety-critical vehicular systems. Extensive evaluation across 4 platoon controllers, multiple attack vectors, and diverse mobility scenarios demonstrates superior performance (<inline-formula> <tex-math>$geq 0.93$ </tex-math></inline-formula>) compared to state-of-the-art baseline architectures. A comprehensive deployment analysis utilizing TensorFlow Lite (TFLite), Open Neural Network Exchange (ONNX), and TensorRT achieves sub-millisecond inference latency, making it suitable for real-time operation on resource-constrained edge platforms. Hence, validating <sc>AIMformer</small> is viable for both in-vehicle and roadside deployment.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"27 10","pages":"13379-13395"},"PeriodicalIF":9.1,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11640362","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097973","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}