检索结果-内蒙古大学图书馆

Informatica (Slovenia) 2025年第10期49卷 85-100页

作者： Gu, Mulan School of Mechanical Engineering Southwest Jiaotong University Sichuan Chengdu611756 China

This study analyzes the basic principles and structural models of multi-sensor data fusion, and emphasizes the importance of effective fusion algorithms. The proposed improved weighted information fusion algorithm combines the Jackknife method and the adaptive weighting method, while the improved Kalman filter fusion algorithm combines the weighting factor, the state transfer matrix, the measurement transfer matrix and the process noise distribution matrix. The effectiveness of the proposed algorithm is verified through extensive simulations and practical applications, and the estimation accuracy is improved by 15% and the error rate is reduced by 20% compared with the traditional methods. The improved algorithm shows stronger robustness in dealing with the uncertainty and inconsistency of sensor data, and the adaptive weighting mechanism effectively solves the problem of sensor reliability variation. The computational complexity and scalability of the proposed algorithms are analyzed, and it is shown that they can be implemented in large-scale sensor networks in real time. This study also discusses the challenges and opportunities of multi-sensor data fusion, including the integration of deep learning techniques, the extension to heterogeneous sensor scenarios, and the need for broader experimental validation in real-world precision measurement applications. © 2025 Slovene Society Informatika. All rights reserved.

关键词： Kalman filters

来源：评论

学校读者我要写书评

暂无评论

Roadside sensor Systems for vulnerable Road User Protection: A Review of Methods and applications

引用

IEEE ACCESS 2025年 13卷 62717-62738页

作者： Zhang, Tianya Cheng, Lei Bang, Tam Guo, Lihao Hajij, Mustafa Cao, Siyang Harris, Austin Sartipi, Mina Univ Tennessee Chattanooga Ctr Urban Informat & Progress CUIP UTC Res Inst Chattanooga TN 37405 USA Univ Arizona Dept Elect & Comp Engn Tucson AZ 85721 USA Univ San Francisco Elect Engn Dept San Francisco CA 94117 USA

This review surveyed infrastructure-based sensor technologies for protecting vulnerable road users (vRUs) in traffic safety systems, focusing on four key components of the roadside system: calibration methods, sensor fusion approaches, trajectory prediction, and risk analysis frameworks. The paper analyzes different sensor types, discussing their characteristics, advantages, and limitations. Key challenges in sensor calibration for roadside deployment are addressed, alongside various fusion strategies at data, feature, and decision levels. The review covers trajectory prediction methods, from classical approaches to deep learning architectures, examining their applications in vRU behavior analysis. A comprehensive evaluation of safety assessment methodologies using surrogate safety measures is provided, considering both single and multi-sensor implementations. Technical challenges including scalability, real-time processing, and sensor synchronization are discussed, while identifying opportunities in emerging technologies such as advanced AI and vehicle-to-everything (v2X) integration. The paper concludes by addressing future research directions and ethical implications for improving urban vRU safety.

关键词： Safety Roads Surveys vehicle-to-everything Calibration Protection Pedestrians Real-time systems Accidents Trajectory vulnerable road users adaptive calibration sensor fusion prediction and safety analysis

来源：评论

学校读者我要写书评

暂无评论

Knowledge Transfer for Body sensor Networks: Characterization and Resistance of Negative Transfer in Overconstrained Environments

引用

IEEE INTERNET OF THINGS JOURNAL 2024年第19期11卷 31959-31973页

作者： Shi, Han Liu, Yang Luo, Hao Zhao, Hai Liaoning Univ Coll Informat Shenyang 110036 Peoples R China Liaoning Univ Sch Phys Shenyang 110036 Peoples R China Northeastern Univ Sch Comp Sci & Engn Shenyang 110169 Peoples R China

Transfer learning (TL) has shown potential for body sensor network (BSN) applications. However, TL performance is influenced by various factors, such as the BSN environment, user-specific physiological signs, or sensor modalities, leading to negative transfer (NT) effects. Limited research has defined NT from perspectives, such as data quality, domain transferability, and transfer components, but the cognition of NT in multimodal BSNs remains insufficient. In this article, we explore the correlation between transfer modes and NT, proposing a knowledge transfer framework composed of common domain adaptation constraints to reveal the NT effects in overconstrained environments. This effect can explain why strongly constrained transfer algorithms are effective but not necessarily optimal in terms of performance. In the three typical BSN scenarios of emotion recognition, fall detection, and daily activity recognition, we demonstrate the NT in overconstrained environments and its potential influencing factors using different constraint combinations and based on different multimodal body sensor fusion architectures. Extensive experiments focus on discussing the impact of category-condition constraints, domain similarity, and fusion architectures on the NT effect in BSNs. This work can provide a new perspective for the design of multimodal BSN knowledge transfer scheme.

关键词： Transfer learning Task analysis Data models Internet of Things Knowledge transfer Accuracy Training Body sensor network (BSN) information fusion negative transfer (NT) transfer learning (TL)

来源：评论

学校读者我要写书评

暂无评论

RGB-LiDAR fusion for accurate 2D and 3D object detection

引用

MACHINE vISION AND applications 2023年第5期34卷 1-11页

作者： Mousa-Pasandi, Morteza Liu, Tianran Massoud, Yahya Laganiere, Robert Univeristy Ottawa Ottawa ON Canada

Effective detection of road objects in diverse environmental conditions is a critical requirement for autonomous driving systems. Multi-modal sensor fusion is a promising approach for improving perception, as it enables the combination of information from multiple sensor streams in order to optimize the integration of their respective data. fusion operators are employed within fully convolutional architectures to combine features derived from different modalities. In this research, we present a framework that utilizes early fusion mechanisms to train and evaluate 2D object detection algorithms. Our evaluation shows that sensor fusion outperforms RGB-only detection methods, yielding a boost of +15.07% for car detection, +10.81% for pedestrian detection, and +19.86% for cyclist detection. In our comparative study, we evaluated three arithmetic-based fusion operators and two learnable fusion operators. Furthermore, we conducted a performance comparison between early- and mid-level fusion techniques and investigated the effects of early fusion on state-of-the-art 3D object detectors. Lastly, we provide a comprehensive analysis of the computational complexity of our proposed framework, along with an ablation study.

关键词： Multimodal deep learning sensor fusion fusion operators Object detection Autonomous driving Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

AiCareGaitRehabilitation: Multi-modalities sensor data fusion for AI-IoT enabled realtime electrical stimulation device for pre-Fog and post-Fog to person with Parkinson's Disease

引用

INFORMATION fusion 2025年 122卷

作者： Ghayvat, Hemant Awais, Muhammad Geddam, Rebakah Quasim, Mohammad Tabrez Khowaja, Sunder Ali Dev, Kapal Linnaeus Univ Fac Technol Comp Sci & Media Technol Dept PG Vejdes Vag 351 95 S-35234 Vaxjo Sweden Mem Sloan Kettering Canc Ctr Dept Med Phys New York NY 10065 USA Karnavati Univ Unitedworld Inst Technol Gandhinagar 382422 Gujarat India Univ Bisha Coll Comp & Informat Technol Dept Comp Sci & Artificial Intelligence Bisha 551 Saudi Arabia Dublin City Univ ADAPT Ctr Dublin D09 V209 Ireland Dublin City Univ Sch Comp Dublin D09 V209 Ireland Munster Technol Univ ADAPT Ctr Bishopstown T12 P928 Cork Ireland Munster Technol Univ Dept Comp Sci Bishopstown T12 P928 Cork Ireland Univ Johannesburg Dept Inst intelligent Syst T12 P928 Auckland Pk Johannesburg South Africa Chitkara Univ Inst Engn & Technol Ctr Res Impact & Outcome Rajpura 140401 Punjab India

The neuromuscular dysfunction known as Freezing of Gait (FoG), which is more prevalent in individuals suffering from Parkinson's Disease (PD), significantly reduces the quality of life and increases their risk of falling. Wearable FoG sensing technologies provide timely biofeedback cues to assist people regain control over their gait. However, the devices being bulky, intrusive, and annoyance of current FoG detection algorithms limit their usability in real-world applications. This study proposes a more efficient approach by integrating the FoG detection fusion algorithm into a Functional Electrical Stimulation (FES) module. The design leverages features with low computational requirements and specialized hardware to minimize the use of physical space and memory. The Convolutional Neural Networks (CNN) approach with SvM output was deployed to classify FoG and non-FoG periods in real-time. Additionally, the study uses CNN algorithms in fusion with data from a triaxial accelerometer, strain sensors, and piezoelectric plantar sensors to test shank-worn FoG detection devices. The study demonstrates that electrical stimulation-based cueing strategies significantly improve gait control and mitigate FoG episodes in people with Parkinson's disease. The AiCareGaitRehabilitation system employs a multi-modal sensor fusion strategy to improve the efficacy of the FES device. Data from various sensors-such as strain sensor, 18 plantar sensors, and four quadriceps sensors-the system provides a holistic view of both pre-freezing of Gait (pre-FOG) and post-freezing of Gait (post-FOG) scenarios. This research aims to improve mobility, reduce fall risks, and eventually improve the quality of life for individuals with Parkinson's disease.

关键词： sensor data fusion Parkinson's disease Freezing of gait Gait analysis Electrical stimulation Deep learning Transformer model

来源：评论

学校读者我要写书评

暂无评论

Comparative Analysis and Implementation of Energy-Quality Scalable Multiply-Accumulate architectures for TinyML applications 31

Comparative Analysis and Implementation of Energy-Quality Sc...

引用

31st Conference on Electronics Circuits and Systems

作者： Pelayo, Kenn Danielle C. Alvarez, Anastacia B. Chua, Adelson N. Galapon, Fredrick Angelo R. Tan, Allen Jason A. Quizon, Lawrence Roman A. Univ Philippines Diliman Elect & Elect Engn Inst Quezon City Philippines

ISBN: (纸本)9798350377217;9798350377200

Recent trends in machine learning algorithms have shifted from computer vision applications to near-sensor intelligent data analytics, such as wireless sensor nodes, which are considered as ultra-low power devices. This shift presents new challenges in energy consumption, throughput, and area budgets. This work implements and optimizes Bit fusion, a 2D-scalable divide-and-conquer (D&C) MAC architecture with precision-scaling from 8 to 2 bits, through bitwidth optimizations and by trading-off complexity overheads that are not aligned with the target MobileNet-v2 model. The designs are synthesized using 22-nm FD-SOI CMOS technology. It was determined that the throughput trade-offs of 1-level architectures in 2bx2b mode for 17.37% less energy per operation and 26.84% more throughput per area would be preferred. Architecture variations were also explored wherein sum-together (ST) architectures outperformed sum-apart (SA) architectures with 21.03% reduction in energy per operation and 33.32% increase in throughput per area.

关键词： ASIC deep neural networks digital accelerators multiply-accumulate MAC precision-scalable circuits energy-quality scalable circuits

来源：评论

学校读者我要写书评

暂无评论

DSP architectures of Covariance Intersection fusion Algorithm for Automotive sensor fusion 14

DSP Architectures of Covariance Intersection Fusion Algorith...

引用

14th International Conference on Computing Communication and Networking Technologies, ICCCNT 2023

作者： Praharshita, D.S.L. Achala, G. Srihari, Pathipati Acharya, U. Shripathi Pardhasaradhi, Bethi India National Institute of Technology Karnataka ECE Dept. India Continental Autonomous Mobility India Pvt Ltd Adas Dept. India

ISBN: (纸本)9798350335095

The data fusion from sensors within the automotive vehicle is vital for improved accuracy and safety. The centralized and information matrix fusion (IMF) algorithms are famous for providing an optimal fusion estimate. However, the IMF is not viable in automotive sensor fusion applications due to the limited bandwidth and low hardware resources. Hence, distributed fusion technology is widely adopted in the automotive sensor applications to achieve high-speed and low-area realizations. This paper proposes three digital signal processing (DSP) architectures for covariance intersection (CI) fusion algorithm: Pipelined-traditional CI, adder-ladder CI, and pipelined adder-ladder CI. The proposed DSP architectures are evaluated with hardware resource consumption (multipliers, adders, and delays), maximum achievable frequency, and latency of the architecture. In addition, proposed CI algorithms for Digital Signal Processing (DSP) architectures are compared with IMF DSP architectures. The hardware resources and optimal pipeline stages required for CI with respect to N number of sensors are provided. The traditional pipeline algorithm requires N number of stages where as the proposed pipelined version of adder-ladder CI requires a N-1 pipeline stage with additional 7N-1 and 7N-3 delay elements for even and odd number of sensors to achieve the overall system operating frequency to an operation of multiplier. The proposed DSP architectures are suitable for automotive sensor fusion due to their high operating frequency and low hardware resources. © 2023 IEEE.

关键词： Digital signal processing

来源：评论

学校读者我要写书评

暂无评论

MULTI-sensor INFORMATION fusion TECHNOLOGY: PRINCIPLE, applications AND KEY TECHNOLOGIES EXPLORATIONS 2

MULTI-SENSOR INFORMATION FUSION TECHNOLOGY: PRINCIPLE, APPLI...

引用

2nd International Conference on Mechatronic Automation and Electrical Engineering, ICMAEE 2024

作者： Wu, Wenchen Xu, Ruizhe Guangdong Country Garden School Foshan528312 China The Affiliated Middle School of Fujian Normal University Fuzhou350007 China

ISBN: (纸本)9781837242672

Multi-sensor information fusion (MSIF) has emerged as a critical technology in various fields due to the growing complexity of systems and the demand for precise information. This paper explores the fundamental principles, key technologies, and potential applications of MSIF. The paper delves into the design and implementation of MSIF systems, discussing various system architectures such as centralized, distributed, and hybrid models. Key technologies crucial to MSIF, including sensor calibration, data association, filtering methods, and fusion algorithms like Bayesian Criterion, Dempster-Shafer Evidence Theory, Fuzzy Theory, and Neural Networks, are thoroughly examined. The study highlights the importance of data preprocessing and feature extraction techniques in achieving efficient data fusion. Furthermore, the potential of MSIF in diagnosing complex systems is demonstrated through applications in industrial automation, environmental monitoring, and healthcare. MSIF also faces challenges related to data synchronization, algorithm complexity, and the integration of diverse data sources. With ongoing advancements in sensor technology and data processing, MSIF is poised to play an increasingly significant role in intelligent systems, including autonomous vehicles and smart cities. © The Institution of Engineering & Technology 2024.

关键词： Information fusion

来源：评论

学校读者我要写书评

暂无评论

Simulating effects of sensor-to-segment alignment errors on IMU-based estimates of lower limb joint angles during running

引用

SPORTS ENGINEERING 2025年第1期28卷 1-11页

作者： Potter, Michael v. Francis Marion Univ Dept Phys & Engn Florence SC 29506 USA

Wearable inertial measurement units offer opportunities to monitor and study running kinematics in relatively unconstrained environments. However, there remain many challenges for accurately estimating joint angles from inertial measurement unit sensor data. One important challenge involves determining the sensor-to-segment alignment parameters which specify the relative positions and orientations between the sensor and anatomical coordinate frames. Errors in these parameters can lead to errors in joint angle estimates, so it is important for practitioners, researchers, and algorithm developers to understand the required accuracy of sensor-to-segment alignment parameters for different applications. In this study, 480,000 simulations were used to investigate the effects of varying levels of simultaneous sensor-to-segment alignment errors on the accuracy of joint angle estimates from an inertial measurement unit-based method for running. The results demonstrate that accurate lower limb joint angle estimates are obtainable with this method when sensor-to-segment alignment errors are low, but these estimates rapidly degrade as errors in the relative orientations between frames grow. The results give guidance on how accurate sensor-to-segment alignment parameters must be for different applications. The methods used in this paper may also provide a valuable framework for assessing the impact of simultaneous sensor-to-segment alignment errors for other inertial measurement unit based algorithms and activities.

关键词： Wearables Kinematics sensor fusion Inertial motion capture

来源：评论

学校读者我要写书评

暂无评论

Optimizing Long-Term Robot Tracking with Multi-Platform sensor fusion

Optimizing Long-Term Robot Tracking with Multi-Platform Sens...

引用

IEEE/CvF Winter Conference on applications of Computer vision (WACv)

作者： Albanese, Giuliano Mitra, Arka Zaech, Jan-Nico Zhao, Yupeng Chhatkuli, Ajad van Gool, Luc Swiss Fed Inst Technol Zurich Switzerland Katholieke Univ Leuven Leuven Belgium INSAIT Sofia Bulgaria

ISBN: (纸本)9798350318920;9798350318937

Monitoring a fleet of robots requires stable long-term tracking with re-identification, which is yet an unsolved challenge in many scenarios. One application of this is the analysis of autonomous robotic soccer games at RoboCup. Tracking in these games requires handling of identically looking players, strong occlusions, and non-professional video recordings, but also offers state information estimated by the robots. In order to make effective use of the information coming from the robot sensors, we propose a robust tracking and identification pipeline. It fuses external non-calibrated camera data with the robots' internal states using quadratic optimization for tracklet matching. The approach is validated using game recordings from previous RoboCup World Cup tournaments.

关键词： algorithms algorithms and algorithms applications formulations Machine learning architectures Robotics video recognition and understanding

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：