IEEE Transactions on Intelligent Transportation Systems最新文献_第8页

Design and Control of Personalized Steering Feel for Steer-by-Wire Systems 线控转向系统个性化转向感觉的设计与控制

IF 7.9 1区工程技术

IEEE Transactions on Intelligent Transportation Systems Pub Date : 2025-04-02 DOI: 10.1109/TITS.2025.3554276

Qingya Zhou;Liang Liu;Zhaoping Xu;Xianhui Wang

{"title":"Design and Control of Personalized Steering Feel for Steer-by-Wire Systems","authors":"Qingya Zhou;Liang Liu;Zhaoping Xu;Xianhui Wang","doi":"10.1109/TITS.2025.3554276","DOIUrl":"https://doi.org/10.1109/TITS.2025.3554276","url":null,"abstract":"The realization of personalized steering feel plays a crucial role in prompting the widespread adoption and driver acceptance of autonomous vehicles (AVs). Since human drivers have distinctly different steering feel preferences and requirements, the personalized matching is essential for ensuring safety and comfort of human-vehicle collaboration. Therefore, this paper proposed a novel personalized steering feel design method based on the driver’s steering characteristics, utilizing steer-by-wire (SBW) vehicle as the platform. Specifically, the personalized steering feel was composed of conventional steering torque, customizable module and mechanical compensation. The conventional steering torque was calculated by a tire model. Then, the concept of road sense style was the first introduced as the basis for the design of the customizable module. Furthermore, a complete road sense style recognition system was put forward from the perspective of semi-supervised learning. The driver’s steering characteristics were collected and classified by k-means algorithm, and a generic road sense style recognition model was built and optimized through support vector machine with crow search (CS-SVM) method. The recognition results were integrated into the design of the personalized steering feel. In addition, a steering controller adopting sliding mode control (SMC) was designed to facilitate stable and rational steering maneuvers by drivers. Finally, driving experiments were performed. The results show that the designed steering feel can rapidly adapt to diverse drivers’ preferences, offering a more satisfying driving experience and improving overall safety.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 5","pages":"6288-6303"},"PeriodicalIF":7.9,"publicationDate":"2025-04-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143908369","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}

引用次数: 0

Forecasting Traffic Progression in Terms of Semantically Interpretable States by Exploring Multiple Data Representations 通过探索多种数据表示，以语义可解释状态预测流量进展

IF 7.9 1区工程技术

IEEE Transactions on Intelligent Transportation Systems Pub Date : 2025-04-02 DOI: 10.1109/TITS.2025.3553238

Michiel Dhont;Adrian Munteanu;Elena Tsiporkova

{"title":"Forecasting Traffic Progression in Terms of Semantically Interpretable States by Exploring Multiple Data Representations","authors":"Michiel Dhont;Adrian Munteanu;Elena Tsiporkova","doi":"10.1109/TITS.2025.3553238","DOIUrl":"https://doi.org/10.1109/TITS.2025.3553238","url":null,"abstract":"In the rapidly evolving landscape of mobility modelling, the application of deep learning approaches introduces both opportunities and challenges. Such approaches, while powerful, yield opaque models that lack interpretability and adaptability to diverse traffic contexts. Addressing those challenges, a finite set of humanly-interpretable traffic states is exploited here for the purpose of facilitating the annotation of mobility data with meaningful labels such as congestion, free-flow, traffic build-up, etc. Such annotation unlocks a range of opportunities to integrate multiple complementary approaches for modelling state transition behaviour. Concretely, a novel hybrid modelling framework is introduced in this article leveraging multiple data representations (temporal, time-frequency and symbolic) with the aim to forecast traffic progression in terms of humanly-explicable state transitions. Three distinct modelling paradigms are subsequently explored: neural, neural-to-symbolic, and symbolic-to-neural, by demonstrating their potential to capture and forecast traffic dynamics on real-world mobility data. While the fully neural approach is undoubtedly the most accurate one, the two neuro-symbolic approaches offer a better trade-off between accuracy on one side and interpretability, probability calibration, and computational efficiency, on the other. This work illustrates the importance of tailored data representations in understanding and predicting complex mobility behaviour, highlighting the benefits of hybrid approaches in achieving interpretability and efficiency in traffic data analysis.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 6","pages":"8368-8381"},"PeriodicalIF":7.9,"publicationDate":"2025-04-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10947544","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144196777","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}

引用次数: 0

HSTI: A Light Hierarchical Spatial-Temporal Interaction Model for Map-Free Trajectory Prediction HSTI：用于无地图轨迹预测的轻型分层时空相互作用模型

IF 7.9 1区工程技术

IEEE Transactions on Intelligent Transportation Systems Pub Date : 2025-04-02 DOI: 10.1109/TITS.2025.3543309

Xiaoyang Luo;Shuaiqi Fu;Baolin Gao;Yanan Zhao;Huachun Tan;Zeye Song

{"title":"HSTI: A Light Hierarchical Spatial-Temporal Interaction Model for Map-Free Trajectory Prediction","authors":"Xiaoyang Luo;Shuaiqi Fu;Baolin Gao;Yanan Zhao;Huachun Tan;Zeye Song","doi":"10.1109/TITS.2025.3543309","DOIUrl":"https://doi.org/10.1109/TITS.2025.3543309","url":null,"abstract":"Trajectory prediction is a crucial task for autonomous driving, but current models’ reliance on high-definition (HD) maps limits their broader applicability. To cope with this challenge, we propose a novel map-free trajectory prediction method that leverages spatiotemporal attention mechanisms. The method consists of three key stages: 1) we first encode spatial and temporal features separately using spatial and temporal attention mechanisms, 2) we then model spatial and temporal interactions through Crystal Graph Convolutional Networks (CGCN) and Multi-Head Attention (MHA), 3) finally, an adaptive anchor generation technique is introduced to tackle the multimodal trajectory prediction challenge. This self-adaptive technique generates context-specific anchors, enabling accurate prediction of multiple possible future vehicle trajectories. Extensive experiments on the Argoverse1 and V2X-Seq datasets validate the effectiveness of our approach. On the Argoverse1 dataset, our method outperforms CRAT-Pred by 5.8% in minADE and 6.25% in minFDE. On the V2X-Seq dataset, it achieves improvements of 82.6%, 85.1%, and 44.0% in minADE, minFDE, and MR, respectively, compared to the baseline model.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 6","pages":"9037-9046"},"PeriodicalIF":7.9,"publicationDate":"2025-04-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144205916","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}

引用次数: 0

Physics-Informed Data-Driven Power Capacity Prediction of Lithium-Ion Battery Against Various Temperatures 不同温度下锂离子电池的物理数据驱动功率容量预测

IF 7.9 1区工程技术

IEEE Transactions on Intelligent Transportation Systems Pub Date : 2025-04-02 DOI: 10.1109/TITS.2025.3549458

Guangxin Gao;Guangzhong Dong;Yunjiang Lou;Li Sun;Jingwen Wei

{"title":"Physics-Informed Data-Driven Power Capacity Prediction of Lithium-Ion Battery Against Various Temperatures","authors":"Guangxin Gao;Guangzhong Dong;Yunjiang Lou;Li Sun;Jingwen Wei","doi":"10.1109/TITS.2025.3549458","DOIUrl":"https://doi.org/10.1109/TITS.2025.3549458","url":null,"abstract":"Lithium-ion batteries are extensively utilized in applications ranging from portable electronics to electric vehicles and renewable energy systems. Accurate prediction of the state of power capacity (SOP) in lithium-ion batteries is fundamental for guaranteeing the safe, reliable, and efficient operation of these systems. However, most existing SOP prediction algorithms only account for the external measurable state constraints of the battery, ignoring the influence of the internal electrochemical states. Using electrochemical models to model batteries can introduce the electrochemical perspective, but many related methods ignore the impact of temperature variations on model parameters. Therefore, this paper proposes an SOP estimation framework based on a physics-informed data-driven approach, which fully integrates the electrochemical model and battery operation data to provide accurate power capacity estimation against temperature effects. First, the battery is modeled using an electrochemical model, and the battery operation data is used to identify the electrochemical temperature-sensitive parameters to enhance the accuracy of the model. Secondly, safety constraints for battery operations are introduced from the perspective of the battery mechanisms and the bisection method is employed to search for the maximum current. Compared with the SOP calibration results and the state-of-the-art method, the results highlight the accuracy of the proposed method. Finally, by referring to the characteristic maps-based method and employing Gaussian process regression, the search interval of SOP is significantly reduced based on historical data, reducing the search time by 80%.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 6","pages":"8670-8681"},"PeriodicalIF":7.9,"publicationDate":"2025-04-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144196619","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}

引用次数: 0

Vision-Based Driving Decision Making Using Multi-Action Deep Q Network 基于视觉的多动作深度Q网络驾驶决策

IF 7.9 1区工程技术

IEEE Transactions on Intelligent Transportation Systems Pub Date : 2025-04-02 DOI: 10.1109/TITS.2025.3552993

Sheng Yuan;Yaochen Li;Kai Zhao;Li Zhu;Jiaxin Guo;Xinnan Ma;Yuncheng Xu

{"title":"Vision-Based Driving Decision Making Using Multi-Action Deep Q Network","authors":"Sheng Yuan;Yaochen Li;Kai Zhao;Li Zhu;Jiaxin Guo;Xinnan Ma;Yuncheng Xu","doi":"10.1109/TITS.2025.3552993","DOIUrl":"https://doi.org/10.1109/TITS.2025.3552993","url":null,"abstract":"The performance improvement of perception algorithms and integrated validation of driving decision making remains challenging in the fields of computer vision and intelligent transportation systems. In this paper, we propose a novel vision-based framework for driving decision making, which is composed of three stages: object perception, lane line perception and driving decision making. For the object perception stage, an improved object perception model named CenterNet-ARA is developed, composing of a new adversarial training method, a receptive field enhancement module and an adaptive sample allocation equalization strategy to fuse multi-scale feature maps. For the lane line perception stage, a lane line perception method named Lite-MobileTR is proposed, which contains an improved Lite-MobileNetV3 encoder and an improved lite-transformer decoder. Moreover, a noise removal task is incorporated to alleviate the problem of slow convergence speed caused by Hungarian loss function. For the driving decision making stage, a new Multi-Action DQN is proposed utilizing a vehicle curriculum learning strategy and a curiosity exploration strategy to alleviate the problem of random exploration in the learning process. The proposed framework is evaluated on the Tusimple, CULane, TSD-max, and KITTI datasets. Finally, an integration verification is performed in Carla simulator to validate the driving decision making process. The experimental results well demonstrate the effectiveness of the proposed framework.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 5","pages":"5816-5831"},"PeriodicalIF":7.9,"publicationDate":"2025-04-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143913590","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}

引用次数: 0

MSSF: A 4D Radar and Camera Fusion Framework With Multi-Stage Sampling for 3D Object Detection in Autonomous Driving 基于多阶段采样的4D雷达与相机融合框架在自动驾驶中的三维目标检测

IF 7.9 1区工程技术

IEEE Transactions on Intelligent Transportation Systems Pub Date : 2025-04-02 DOI: 10.1109/TITS.2025.3554313

Hongsi Liu;Jun Liu;Guangfeng Jiang;Xin Jin

{"title":"MSSF: A 4D Radar and Camera Fusion Framework With Multi-Stage Sampling for 3D Object Detection in Autonomous Driving","authors":"Hongsi Liu;Jun Liu;Guangfeng Jiang;Xin Jin","doi":"10.1109/TITS.2025.3554313","DOIUrl":"https://doi.org/10.1109/TITS.2025.3554313","url":null,"abstract":"As one of the automotive sensors that have emerged in recent years, 4D millimeter-wave radar has a higher resolution than conventional 3D radar and provides precise elevation measurements. But its point clouds are still sparse and noisy, making it challenging to meet the requirements of autonomous driving. Camera, as another commonly used sensor, can capture rich semantic information. As a result, the fusion of 4D radar and camera can provide an affordable and robust perception solution for autonomous driving systems. However, previous radar-camera fusion methods have not yet been thoroughly investigated, resulting in a large performance gap compared to LiDAR-based methods. Specifically, they ignore the feature-blurring problem and do not deeply interact with image semantic information. To this end, we present a simple but effective multi-stage sampling fusion (MSSF) network based on 4D radar and camera. On the one hand, we design a fusion block that can deeply interact point cloud features with image features, and can be applied to commonly used single-modal backbones in a plug-and-play manner. The fusion block encompasses two types, namely, simple feature fusion (SFF) and multi-scale deformable feature fusion (MSDFF). The SFF is easy to implement, while the MSDFF has stronger fusion abilities. On the other hand, we propose a semantic-guided head to perform foreground-background segmentation on voxels with voxel feature re-weighting, further alleviating the problem of feature blurring. Extensive experiments on the View-of-Delft (VoD) and TJ4DRadset datasets demonstrate the effectiveness of our MSSF. Notably, compared to state-of-the-art methods, MSSF achieves a 7.0% and 4.0% improvement in 3D mean average precision on the VoD and TJ4DRadSet datasets, respectively. It even surpasses classical LiDAR-based methods on the VoD dataset.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 6","pages":"8641-8656"},"PeriodicalIF":7.9,"publicationDate":"2025-04-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144196937","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}

引用次数: 0

Real-Time Intelligent Landing-Management Under Urban Unpredictable Operations 城市不可预测运行下的实时智能着陆管理

IF 7.9 1区工程技术

IEEE Transactions on Intelligent Transportation Systems Pub Date : 2025-04-02 DOI: 10.1109/TITS.2025.3554453

Behzad Baleghi;Seyed Mohammad-Bagher Malaek

{"title":"Real-Time Intelligent Landing-Management Under Urban Unpredictable Operations","authors":"Behzad Baleghi;Seyed Mohammad-Bagher Malaek","doi":"10.1109/TITS.2025.3554453","DOIUrl":"https://doi.org/10.1109/TITS.2025.3554453","url":null,"abstract":"The integration of electric Vertical Takeoff and Landing (eVTOL) vehicles into urban transportation presents challenges in scalability, real-time adaptability, and operational efficiency. This study introduces an Agent-Based Modeling (ABM) framework for traffic management, dynamically assigning eVTOLs to landing pads based on real-time data to enhance efficiency in congested urban environments. A comparative analysis is conducted across various scheduling and sequencing approaches, including Mixed-Integer Linear Programming (MILP), Time-Advance (TA), heuristic methods, receding horizon scheduling, reinforcement learning (RL)-based frameworks, and decentralized agent-based strategies. While MILP and TA offer structured scheduling, they struggle with scalability. Heuristic and receding horizon methods improve adaptability but require frequent recomputation, and RL-based approaches show promise but demand extensive training. Current Decentralized models support distributed decision-making but face efficiency constraints at scale. The proposed ABM framework effectively manages 200 eVTOLs with near-linear computational scaling, facilitating real-time negotiations and reducing computational bottlenecks seen in centralized models. Simulation results indicate improved assignment efficiency, landing pad utilization, and reduced negotiation times under high-density conditions. As UAM systems expand, ABM may contribute to operational resilience. Future work will explore integrating environmental factors to further enhance robustness.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 6","pages":"8247-8256"},"PeriodicalIF":7.9,"publicationDate":"2025-04-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144196607","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}

引用次数: 0

Pedestrian–Vehicle Interaction Analysis Based on Concept of Dynamic Straight-Right Lane at Signalized Intersection 基于信号交叉口动态直右车道概念的人-车交互分析

IF 7.9 1区工程技术

IEEE Transactions on Intelligent Transportation Systems Pub Date : 2025-04-02 DOI: 10.1109/TITS.2025.3548724

Shidong Liang;Tianyu Zhao;Jing Zhao

{"title":"Pedestrian–Vehicle Interaction Analysis Based on Concept of Dynamic Straight-Right Lane at Signalized Intersection","authors":"Shidong Liang;Tianyu Zhao;Jing Zhao","doi":"10.1109/TITS.2025.3548724","DOIUrl":"https://doi.org/10.1109/TITS.2025.3548724","url":null,"abstract":"Straight-right mixed lanes are common in urban signalized intersections. Right-turning vehicles during green light will conflict with pedestrians, increasing the risk to pedestrian safety. In addition, right-turning vehicles cannot pass the intersection during the red light, which seriously affects the saturation rate of the traffic flow. To solve the above problems, this paper proposes dynamic straight-right lane (DSRL) design scheme that is able to separate straight-through and right-turning vehicles in time and space. Based on the driving characteristics of right-turning vehicles, the operating rules for DSRL have been established. Combined with the intersection design, DSRL control strategy is proposed under the linkage of intersection traffic light and pre-signals. By studying the pedestrian crossing characteristics, the conflict between right-turning vehicles entering the intersection and crossing pedestrians is analyzed under the DSRL. Vehicle delays and traffic capacity are quantified under the DSRL on the basis of vehicle operating rules. The vehicle delay model is tested using SUMO simulation software to prove the validity of the model proposed in this paper. Finally, the simulation environment based on real intersections was set up using MATLAB and sensitivity analyses of the main impact parameters were carried out. It has been experimentally demonstrated that the DSRL not only effectively reduces the delay of both straight-through and right-turning vehicles, but also improves the actual capacity of the lane. In addition, the DSRL reduces conflicts between right-turning vehicles and pedestrians crossing the street. The experimental results can effectively prove that the DSRL can improve the traffic efficiency of vehicle and ensure the safety of pedestrians.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 5","pages":"6017-6041"},"PeriodicalIF":7.9,"publicationDate":"2025-04-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143913306","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}

引用次数: 0

Deep Reinforcement Learning-Based Computation Computational Offloading for Space–Air–Ground Integrated Vehicle Networks 基于深度强化学习的天空地一体化车辆网络计算卸载

IF 7.9 1区工程技术

IEEE Transactions on Intelligent Transportation Systems Pub Date : 2025-04-02 DOI: 10.1109/TITS.2025.3551636

Wenxuan Xie;Chen Chen;Ying Ju;Jun Shen;Qingqi Pei;Houbing Song

{"title":"Deep Reinforcement Learning-Based Computation Computational Offloading for Space–Air–Ground Integrated Vehicle Networks","authors":"Wenxuan Xie;Chen Chen;Ying Ju;Jun Shen;Qingqi Pei;Houbing Song","doi":"10.1109/TITS.2025.3551636","DOIUrl":"https://doi.org/10.1109/TITS.2025.3551636","url":null,"abstract":"In remote or disaster areas, where terrestrial networks are difficult to cover and Terrestrial Edge Computing (TEC) infrastructures are unavailable, solving the computation computational offloading for Internet of Vehicles (IoV) scenarios is challenging. Current terrestrial networks have high data rates, great connectivity, and low delay, but global coverage is limited. Space–Air–Ground Integrated Networks (SAGIN) can improve the coverage limitations of terrestrial networks and enhance disaster resistance. However, the rising complexity and heterogeneity of networks make it difficult to find a robust and intelligent computational offload strategy. Therefore, joint scheduling of space, air, and ground resources is needed to meet the growing demand for services. In light of this, we propose an integrated network framework for Space-Air Auxiliary Vehicle Computation (SA-AVC) and build a system model to support various IoV services in remote areas. Our model aims to maximize delay and fair utility and increase the utilization of satellites and Autonomous aerial vehicles (AAVs). To this end, we propose a Deep Reinforcement Learning algorithm to achieve real-time computational computational offloading decisions. We utilize the Rank-based Prioritization method in Prioritized Experience Replay (PER) to optimize our algorithm. We designed simulation experiments for validation and the results show that our proposed algorithm reduces the average system delay by 17.84%, 58.09%, and 58.32%, and the average variance of the task completion delay will be reduced by 29.41%, 48.74%, and 49.58% compared to the Deep Q Network (DQN), Q-learning and RandomChoose algorithms.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 5","pages":"5804-5815"},"PeriodicalIF":7.9,"publicationDate":"2025-04-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143913497","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}

引用次数: 0

TSceneJAL: Joint Active Learning of Traffic Scenes for 3D Object Detection TSceneJAL：面向三维目标检测的交通场景联合主动学习

IF 7.9 1区工程技术

IEEE Transactions on Intelligent Transportation Systems Pub Date : 2025-04-01 DOI: 10.1109/TITS.2025.3553170

Chenyang Lei;Weiyuan Peng;Guang Zhou;Meiying Zhang;Qi Hao;Chunlin Ji;Chengzhong Xu

{"title":"TSceneJAL: Joint Active Learning of Traffic Scenes for 3D Object Detection","authors":"Chenyang Lei;Weiyuan Peng;Guang Zhou;Meiying Zhang;Qi Hao;Chunlin Ji;Chengzhong Xu","doi":"10.1109/TITS.2025.3553170","DOIUrl":"https://doi.org/10.1109/TITS.2025.3553170","url":null,"abstract":"Most autonomous driving (AD) datasets incur substantial costs for collection and labeling, inevitably yielding a plethora of low-quality and redundant data instances, thereby compromising performance and efficiency. Many applications in AD systems necessitate high-quality training datasets using both existing datasets and newly collected data. In this paper, we propose a traffic scene joint active learning (TSceneJAL) framework that can efficiently sample the balanced, diverse, and complex traffic scenes from both labeled and unlabeled data. The novelty of this framework is threefold: 1) a scene sampling scheme based on a category entropy, to identify scenes containing multiple object classes, thus mitigating class imbalance for the active learner; 2) a similarity sampling scheme, estimated through the directed graph representation and a marginalize kernel algorithm, to pick sparse and diverse scenes; 3) an uncertainty sampling scheme, predicted by a mixture density network, to select instances with the most unclear or complex regression outcomes for the learner. Finally, the integration of these three schemes in a joint selection strategy yields an optimal and valuable subdataset. Experiments on the KITTI, Lyft, nuScenes and SUScape datasets demonstrate that our approach outperforms existing state-of-the-art methods on 3D object detection tasks with up to 12% improvements.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 6","pages":"8424-8440"},"PeriodicalIF":7.9,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144196740","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}

引用次数: 0