Measurement最新文献_第2页

Research on the error avoidance compensation strategy of a kinematically redundant parallel mechanism based on the error model 基于误差模型的运动冗余并联机构误差避免补偿策略研究

IF 5.6 2区工程技术 Q1 ENGINEERING, MULTIDISCIPLINARY

Measurement

Pub Date : 2026-02-07 DOI: 10.1016/j.measurement.2026.120725

Buqin Hu , Jiamei Wang , Haibo Qu , Zhizhen Zhou , Sheng Guo

The positioning accuracy of parallel mechanisms is critical in advanced industrial applications. This paper presents an error avoidance and compensation strategy based on an error model, using a spatial 1T2R type three-degree-of-freedom (3-DOF) kinematically redundant parallel mechanism (KR-PM) as an example. First, a generalized method for establishing a mechanism error model is introduced using matrix differentiation. Error sensitivity evaluation indices that reflect the actual error transmission relationships are also defined. Applying this method, the inverse kinematics and error models of the spatial 3PRR(RR)S-P KR-PM are developed, where P denotes a prismatic joint, R a revolute joint, S a spherical joint, and an underlined letter indicates an actuated joint. Subsequently, an error avoidance strategy is proposed based on error sensitivity indices and static error (i.e., time-invariant manufacturing and assembly errors) to mitigate their impact on positioning accuracy. Furthermore, an error compensation strategy incorporating a spring element is introduced to counteract the influence of joint clearance errors. Theoretical and simulation results demonstrate that the proposed error avoidance and compensation strategies effectively enhance the positioning accuracy of the mechanism when executing task paths.

在先进的工业应用中，并联机构的定位精度至关重要。以空间1T2R型三自由度冗余并联机构为例，提出了一种基于误差模型的误差避免与补偿策略。首先，介绍了一种利用矩阵微分法建立机构误差模型的广义方法。定义了反映实际误差传递关系的误差灵敏度评价指标。应用该方法，建立了空间3PRR(RR)S-P KR-PM的运动学逆模型和误差模型，其中P为移动关节，R为转动关节，S为球面关节，带下划线的字母表示驱动关节。随后，提出了一种基于误差敏感性指标和静态误差（即定常制造和装配误差）的误差避免策略，以减轻它们对定位精度的影响。此外，引入了一种包含弹簧元件的误差补偿策略来抵消关节间隙误差的影响。理论和仿真结果表明，所提出的误差避免和补偿策略有效地提高了机构在执行任务路径时的定位精度。

{"title":"Research on the error avoidance compensation strategy of a kinematically redundant parallel mechanism based on the error model","authors":"Buqin Hu , Jiamei Wang , Haibo Qu , Zhizhen Zhou , Sheng Guo","doi":"10.1016/j.measurement.2026.120725","DOIUrl":"10.1016/j.measurement.2026.120725","url":null,"abstract":"<div><div>The positioning accuracy of parallel mechanisms is critical in advanced industrial applications. This paper presents an error avoidance and compensation strategy based on an error model, using a spatial 1T2R type three-degree-of-freedom (3-DOF) kinematically redundant parallel mechanism (KR-PM) as an example. First, a generalized method for establishing a mechanism error model is introduced using matrix differentiation. Error sensitivity evaluation indices that reflect the actual error transmission relationships are also defined. Applying this method, the inverse kinematics and error models of the spatial 3<u>P</u>RR(RR)S-<u>P</u> KR-PM are developed, where P denotes a prismatic joint, R a revolute joint, S a spherical joint, and an underlined letter indicates an actuated joint. Subsequently, an error avoidance strategy is proposed based on error sensitivity indices and static error (i.e., time-invariant manufacturing and assembly errors) to mitigate their impact on positioning accuracy. Furthermore, an error compensation strategy incorporating a spring element is introduced to counteract the influence of joint clearance errors. Theoretical and simulation results demonstrate that the proposed error avoidance and compensation strategies effectively enhance the positioning accuracy of the mechanism when executing task paths.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"269 ","pages":"Article 120725"},"PeriodicalIF":5.6,"publicationDate":"2026-02-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146147472","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

High accuracy sitting posture monitoring enabled by multiplexing fiber optic microsphere array and deep learning 通过多路光纤微球阵列和深度学习实现高精度坐姿监测

IF 5.6 2区工程技术 Q1 ENGINEERING, MULTIDISCIPLINARY

Measurement

Pub Date : 2026-02-06 DOI: 10.1016/j.measurement.2026.120710

Jianwei Li , Meiqi Gao , Jialin Li , Ying Xu , Wanli Tian , Zhongping Zhang , Jiuru Yang

In this paper, we presents a microsphere-based fiber-optic sensor array designed for sitting posture monitoring. The sensor features a hollow-core fiber with silica walls forming an air cavity, enabling high-resolution pressure sensing through intensity demodulation. Encapsulation with polydimethylsiloxane enhances the durability and compatibility of sensors with human interfaces. Comprehensive theoretical analysis and experimental validation identify the optimal cavity length of 48 μm and achieve a wavelength sensitivity of 0.15 nm/N with the range of 0 − 21 N. A six-sensor array is positioned on a chair and precise pressure distribution detection is achieved by the technique of time-division multiplexing. Key pressure areas associated with eight distinct sitting postures are precisely and rapidly captured. Combined with a machine learning model, the system achieves an accuracy of 94.55 % in posture recognition. Testing on previously unseen data achieves an accuracy of 92.25 %, further evaluating the model’s generalization capability.

本文提出了一种基于微球的光纤传感器阵列，用于坐姿监测。该传感器具有中空芯光纤，其硅壁形成空腔，通过强度解调实现高分辨率压力传感。聚二甲基硅氧烷封装增强了传感器与人机界面的耐用性和兼容性。综合理论分析和实验验证，确定了最佳空腔长度为48 μm，波长灵敏度为0.15 nm/N，范围为0 ~ 21 N。将六传感器阵列放置在椅子上，采用分时复用技术实现了精确的压力分布检测。与八种不同坐姿相关的关键压力区域被精确而快速地捕获。结合机器学习模型，该系统的姿态识别准确率达到94.55%。对以前未见过的数据进行测试，准确率达到92.25%，进一步评估了模型的泛化能力。

{"title":"High accuracy sitting posture monitoring enabled by multiplexing fiber optic microsphere array and deep learning","authors":"Jianwei Li , Meiqi Gao , Jialin Li , Ying Xu , Wanli Tian , Zhongping Zhang , Jiuru Yang","doi":"10.1016/j.measurement.2026.120710","DOIUrl":"10.1016/j.measurement.2026.120710","url":null,"abstract":"<div><div>In this paper, we presents a microsphere-based fiber-optic sensor array designed for sitting posture monitoring. The sensor features a hollow-core fiber with silica walls forming an air cavity, enabling high-resolution pressure sensing through intensity demodulation. Encapsulation with polydimethylsiloxane enhances the durability and compatibility of sensors with human interfaces. Comprehensive theoretical analysis and experimental validation identify the optimal cavity length of 48 μm and achieve a wavelength sensitivity of 0.15 nm/N with the range of 0 − 21 N. A six-sensor array is positioned on a chair and precise pressure distribution detection is achieved by the technique of time-division multiplexing. Key pressure areas associated with eight distinct sitting postures are precisely and rapidly captured. Combined with a machine learning model, the system achieves an accuracy of 94.55 % in posture recognition. Testing on previously unseen data achieves an accuracy of 92.25 %, further evaluating the model’s generalization capability.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"268 ","pages":"Article 120710"},"PeriodicalIF":5.6,"publicationDate":"2026-02-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146171997","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

Signal processing method to coherent noise suppression with application to enhance measurement quality in laser ultrasound 信号处理方法对相干噪声进行抑制，以提高激光超声测量质量

IF 5.6 2区工程技术 Q1 ENGINEERING, MULTIDISCIPLINARY

Measurement

Pub Date : 2026-02-06 DOI: 10.1016/j.measurement.2026.120749

Patrycja Pyzik, Lukasz Ambrozinski

A laser ultrasound (LU) measurement technique is presented for non-contact inspection of metallic structures, addressing the problem of coherent noise that limits the accuracy of conventional ultrasonic measurements. The proposed method introduces a novel compensation procedure based on experimentally acquired data, enabling effective suppression of deterministic wave components originating from multimodal character of laser excitation. Combined with Frequency-domain Synthetic Aperture Focusing Technique (F-SAFT), the method achieved a measurement accuracy of 0.13 mm, the ability to detect

⌀

0.6 mm flaws with spatial resolution down to 1.1 mm, and a significant improvement in signal-to-noise ratio compared with raw data. The approach was validated on a real engineering structure — a pre-manufactured tailor welded blank. The result demonstrated that the developed technique enhances the interpretability of LU scans and enables clear flaw detection using automatic gating techniques—a task previously impossible with raw images. The proposed method is general and can be applied to various non-destructive testing (NDT) configurations affected by coherent noise.

提出了一种用于金属结构非接触检测的激光超声测量技术，解决了传统超声测量中存在的相干噪声问题。该方法引入了一种基于实验数据的补偿方法，能够有效地抑制由激光激发的多模态特性引起的确定性波分量。结合频域合成孔径聚焦技术（F-SAFT），该方法实现了0.13 mm的测量精度，能够检测出0.6 mm的缺陷，空间分辨率低至1.1 mm，与原始数据相比，信噪比显著提高。该方法在一个实际的工程结构上得到了验证——一个预制的定制焊接坯。结果表明，所开发的技术增强了LU扫描的可解释性，并能够使用自动门控技术进行清晰的缺陷检测-这是以前使用原始图像无法完成的任务。该方法具有通用性，可适用于受相干噪声影响的各种无损检测结构。

{"title":"Signal processing method to coherent noise suppression with application to enhance measurement quality in laser ultrasound","authors":"Patrycja Pyzik, Lukasz Ambrozinski","doi":"10.1016/j.measurement.2026.120749","DOIUrl":"10.1016/j.measurement.2026.120749","url":null,"abstract":"<div><div>A laser ultrasound (LU) measurement technique is presented for non-contact inspection of metallic structures, addressing the problem of coherent noise that limits the accuracy of conventional ultrasonic measurements. The proposed method introduces a novel compensation procedure based on experimentally acquired data, enabling effective suppression of deterministic wave components originating from multimodal character of laser excitation. Combined with Frequency-domain Synthetic Aperture Focusing Technique (F-SAFT), the method achieved a measurement accuracy of 0.13 mm, the ability to detect <span><math><mo>⌀</mo></math></span>0.6 mm flaws with spatial resolution down to 1.1 mm, and a significant improvement in signal-to-noise ratio compared with raw data. The approach was validated on a real engineering structure — a pre-manufactured tailor welded blank. The result demonstrated that the developed technique enhances the interpretability of LU scans and enables clear flaw detection using automatic gating techniques—a task previously impossible with raw images. The proposed method is general and can be applied to various non-destructive testing (NDT) configurations affected by coherent noise.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"269 ","pages":"Article 120749"},"PeriodicalIF":5.6,"publicationDate":"2026-02-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146147307","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

Baseline-free direct laser absorption spectroscopy with sinusoidal modulation: a theoretical and experimental study for combustion diagnostics 无基线直接激光吸收光谱与正弦调制：燃烧诊断的理论和实验研究

IF 5.6 2区工程技术 Q1 ENGINEERING, MULTIDISCIPLINARY

Measurement

Pub Date : 2026-02-06 DOI: 10.1016/j.measurement.2026.120756

Shaojie Wang, Weifan Hu, Shengming Yin, Liangliang Xu, Mingming Gu, Fei Qi

Direct absorption spectroscopy (DAS) typically employs sawtooth waveform modulation, which provides a simple baseline that can be approximated using polynomial fitting. However, at high modulation frequencies, sawtooth modulation imposes demanding hardware requirements and may lead to baseline distortion. To address these limitations, we propose a baseline-free DAS method based on sinusoidal modulation. This approach exploits the distinct frequency-domain amplitude characteristics of sinusoidal modulation. By applying a Fourier transform to the logarithm of the transmitted light intensity, the method decouples absorbance from baseline interference. Moreover, tuning the ratio of the modulation signal’s direct current (DC) and alternating current (AC) components provides additional control over baseline influence. For experimental validation, a laser operating at the bandhead region (4.172 um) of the CO₂ absorption spectrum was employed. In this spectral range, densely packed absorption lines leave virtually non-absorption intervals, rendering conventional baseline fitting methods inaccurate, particularly for temperature measurements. In contrast, the proposed method demonstrated robust and accurate performance. The accuracy and critical parameters of the method were systematically examined through numerical simulations and experimentally validated in a flat flame established over a Hencken burner.

直接吸收光谱（DAS）通常采用锯齿波形调制，它提供了一个简单的基线，可以使用多项式拟合近似。然而，在高调制频率下，锯齿调制施加了苛刻的硬件要求，并可能导致基线失真。为了解决这些限制，我们提出了一种基于正弦调制的无基线DAS方法。这种方法利用了正弦调制独特的频域幅度特性。通过对透射光强度的对数进行傅里叶变换，该方法将吸光度与基线干涉解耦。此外，调整调制信号的直流（DC）和交流（AC）分量的比例提供了对基线影响的额外控制。为了进行实验验证，采用了工作在CO2吸收光谱带头区域（4.172 um）的激光器。在这个光谱范围内，密集排列的吸收线实际上留下了非吸收间隔，使得传统的基线拟合方法不准确，特别是对于温度测量。结果表明，该方法具有较好的鲁棒性和准确性。通过数值模拟系统地检验了该方法的准确性和关键参数，并在henken燃烧器上建立了平坦火焰实验验证了该方法。

{"title":"Baseline-free direct laser absorption spectroscopy with sinusoidal modulation: a theoretical and experimental study for combustion diagnostics","authors":"Shaojie Wang, Weifan Hu, Shengming Yin, Liangliang Xu, Mingming Gu, Fei Qi","doi":"10.1016/j.measurement.2026.120756","DOIUrl":"10.1016/j.measurement.2026.120756","url":null,"abstract":"<div><div>Direct absorption spectroscopy (DAS) typically employs sawtooth waveform modulation, which provides a simple baseline that can be approximated using polynomial fitting. However, at high modulation frequencies, sawtooth modulation imposes demanding hardware requirements and may lead to baseline distortion. To address these limitations, we propose a baseline-free DAS method based on sinusoidal modulation. This approach exploits the distinct frequency-domain amplitude characteristics of sinusoidal modulation. By applying a Fourier transform to the logarithm of the transmitted light intensity, the method decouples absorbance from baseline interference. Moreover, tuning the ratio of the modulation signal’s direct current (DC) and alternating current (AC) components provides additional control over baseline influence. For experimental validation, a laser operating at the bandhead region (4.172 um) of the CO<sub>2</sub> absorption spectrum was employed. In this spectral range, densely packed absorption lines leave virtually non-absorption intervals, rendering conventional baseline fitting methods inaccurate, particularly for temperature measurements. In contrast, the proposed method demonstrated robust and accurate performance. The accuracy and critical parameters of the method were systematically examined through numerical simulations and experimentally validated in a flat flame established over a Hencken burner.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"269 ","pages":"Article 120756"},"PeriodicalIF":5.6,"publicationDate":"2026-02-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146147312","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

Gradient Pore-Engineered biocompatible and wearable sensor for real-time leaf moisture monitoring 梯度孔工程生物兼容和可穿戴传感器，用于实时监测叶片湿度

IF 5.6 2区工程技术 Q1 ENGINEERING, MULTIDISCIPLINARY

Measurement

Pub Date : 2026-02-06 DOI: 10.1016/j.measurement.2026.120708

Yi Tian , Xiaoqi Zhou , Juntao Zhu , Mingfu Xiao , Shouyong Xie , Jie Huang , Gang Liu , Yuanyuan Huang

The real-time monitoring of leaf moisture is crucial for understanding the mechanism of physiological activity, breeding crops with drought or flood tolerance, and diagnosing crop physiology status. However, current monitoring methods fail to meet the requirements for in situ, non-destructive monitoring due to limitations in permeability and ductility. Here, we present a biocompatible and wearable sensor to realize real-time monitoring leaf moisture. The sensor incorporates a dual-layer gradient porous structure that forms a vertical export channel, achieving a high moisture evaporation rate (approximately 0.28kg m^-2h⁻¹) and enabling the precise recording of moisture signals and long-term wear. Based on the biocompatible and wearable sensor, we utilized a wireless system to enable continuous real-time moisture monitoring. This approach offers a powerful tool for analyzing key physiological signals in plants and holds potential for adaptation to the real-time monitoring of signaling in other crops.

叶片水分的实时监测对了解作物生理活动机制、选育耐旱或耐涝作物、诊断作物生理状况具有重要意义。然而，目前的监测方法由于渗透性和延性的限制，不能满足现场无损监测的要求。在这里，我们提出了一种生物兼容的可穿戴传感器来实现叶片湿度的实时监测。该传感器采用双层梯度多孔结构，形成垂直出口通道，实现高水分蒸发速率（约0.28kg m-2h−1），并能够精确记录水分信号和长期磨损。基于生物相容性和可穿戴传感器，我们利用无线系统实现连续实时湿度监测。这种方法为分析植物的关键生理信号提供了强大的工具，并具有适应其他作物信号实时监测的潜力。

{"title":"Gradient Pore-Engineered biocompatible and wearable sensor for real-time leaf moisture monitoring","authors":"Yi Tian , Xiaoqi Zhou , Juntao Zhu , Mingfu Xiao , Shouyong Xie , Jie Huang , Gang Liu , Yuanyuan Huang","doi":"10.1016/j.measurement.2026.120708","DOIUrl":"10.1016/j.measurement.2026.120708","url":null,"abstract":"<div><div>The real-time monitoring of leaf moisture is crucial for understanding the mechanism of physiological activity, breeding crops with drought or flood tolerance, and diagnosing crop physiology status. However, current monitoring methods fail to meet the requirements for in situ, non-destructive monitoring due to limitations in permeability and ductility. Here, we present a biocompatible and wearable sensor to realize real-time monitoring leaf moisture. The sensor incorporates a dual-layer gradient porous structure that forms a vertical export channel, achieving a high moisture evaporation rate (approximately 0.28kg m<sup>-2</sup>h<sup>−1</sup>) and enabling the precise recording of moisture signals and long-term wear. Based on the biocompatible and wearable sensor, we utilized a wireless system to enable continuous real-time moisture monitoring. This approach offers a powerful tool for analyzing key physiological signals in plants and holds potential for adaptation to the real-time monitoring of signaling in other crops.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"269 ","pages":"Article 120708"},"PeriodicalIF":5.6,"publicationDate":"2026-02-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146147474","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

Enhanced indoor localization with sign-aided visual–inertial odometry 用符号辅助视觉惯性里程计增强室内定位

IF 5.6 2区工程技术 Q1 ENGINEERING, MULTIDISCIPLINARY

Measurement

Pub Date : 2026-02-06 DOI: 10.1016/j.measurement.2026.120748

Jinle Xu, Wang Gao, Shaopeng He, Hong Liu, Hui Jiang, Shuguo Pan

Visual–Inertial Simultaneous Localization and Mapping (VISLAM) has gained widespread adoption owing to its lightweight architecture, particularly in Global Navigation Satellite System (GNSS) denied indoor spaces. However, the indoor environment typically exhibits textureless surfaces and repetitive structure, which results in sparse and ambiguous features. Such visual degradation can lead to failures in tracking and loop closure detection, causing rapidly accumulating SLAM drift without effective correction mechanisms. This paper presents a sign-aided localization method that identifies signs through vision models and extracts key edge points based on geometric shape-based projection. The landmark coordinate system can then be reconstructed for data association, global relocalization, and drift reduction. By incorporating pre-established sign maps which are efficient and consume little memory, we conducted experiments across three collaborative scenarios, relying only on a monocular camera and an inertial measurement unit (IMU). Experimental results demonstrate that our method outperforms several state-of-the-art (SOTA) VISLAM algorithms in terms of localization accuracy and robustness, significantly reducing accumulated drift. Furthermore, compared to marker-based SLAM methods that rely on artificial landmarks such as ArUco tags and texts, our approach can achieve comparable or superior performance by leveraging naturally occurring signs within the scene, without the need for artificial markers.

视觉惯性同步定位和制图（VISLAM）由于其轻量级架构而获得广泛采用，特别是在全球导航卫星系统（GNSS）的室内空间。然而，室内环境通常呈现无纹理的表面和重复的结构，这导致稀疏和模糊的特征。这种视觉退化会导致跟踪和闭环检测失败，导致SLAM漂移迅速积累，没有有效的纠正机制。本文提出了一种基于视觉模型识别符号，并基于几何形状投影提取关键边缘点的符号辅助定位方法。然后重建地标坐标系统，用于数据关联、全局重定位和减少漂移。通过整合高效且占用内存少的预先建立的标志地图，我们仅依靠单目相机和惯性测量单元（IMU）在三种协作场景下进行了实验。实验结果表明，该方法在定位精度和鲁棒性方面优于几种最先进的（SOTA） VISLAM算法，显著减少了累积漂移。此外，与依赖人工地标（如ArUco标签和文本）的基于标记的SLAM方法相比，我们的方法可以通过利用场景中自然发生的标记来实现相当或更好的性能，而不需要人工标记。

{"title":"Enhanced indoor localization with sign-aided visual–inertial odometry","authors":"Jinle Xu, Wang Gao, Shaopeng He, Hong Liu, Hui Jiang, Shuguo Pan","doi":"10.1016/j.measurement.2026.120748","DOIUrl":"10.1016/j.measurement.2026.120748","url":null,"abstract":"<div><div>Visual–Inertial Simultaneous Localization and Mapping (VISLAM) has gained widespread adoption owing to its lightweight architecture, particularly in Global Navigation Satellite System (GNSS) denied indoor spaces. However, the indoor environment typically exhibits textureless surfaces and repetitive structure, which results in sparse and ambiguous features. Such visual degradation can lead to failures in tracking and loop closure detection, causing rapidly accumulating SLAM drift without effective correction mechanisms. This paper presents a sign-aided localization method that identifies signs through vision models and extracts key edge points based on geometric shape-based projection. The landmark coordinate system can then be reconstructed for data association, global relocalization, and drift reduction. By incorporating pre-established sign maps which are efficient and consume little memory, we conducted experiments across three collaborative scenarios, relying only on a monocular camera and an inertial measurement unit (IMU). Experimental results demonstrate that our method outperforms several state-of-the-art (SOTA) VISLAM algorithms in terms of localization accuracy and robustness, significantly reducing accumulated drift. Furthermore, compared to marker-based SLAM methods that rely on artificial landmarks such as ArUco tags and texts, our approach can achieve comparable or superior performance by leveraging naturally occurring signs within the scene, without the need for artificial markers.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"268 ","pages":"Article 120748"},"PeriodicalIF":5.6,"publicationDate":"2026-02-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146172082","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

LUB-YOLO: A lightweight method for laboratory unsafe behavior detection based on improved YOLOv8s LUB-YOLO：基于改进的YOLOv8s的实验室不安全行为检测轻量级方法

IF 5.6 2区工程技术 Q1 ENGINEERING, MULTIDISCIPLINARY

Measurement

Pub Date : 2026-02-06 DOI: 10.1016/j.measurement.2026.120623

Yulong Chen, Jiangming Kan, Chunjiang Yu

Human unsafe behavior is the main cause of laboratory accidents; however, existing laboratory monitoring methods perform poorly in the face of unsafe behavior. How to prevent unsafe behavior accurately and in real-time when it occurs is considered a key challenge. In this paper, we propose a Laboratory Unsafe Behavior Detection-YOLO model (LUB-YOLO) based on the improved YOLOv8s, which solves the problems of low detection performance and difficulty in distinguishing unsafe behavior. The model combines the Ghost Conv module with HGNet to optimize the computational complexity of the model to improve the detection performance, incorporates the AIFI multi-head attention mechanism to enhance the feature representation and the ability to combine the contextual features, and the dynamic attention mechanism module of DyHead is applied to the detection head to obtain the scale information of the target. The experimental results show that compared with the original network, the LUB-YOLO model improves the [email protected] by 2.1%, reduces the number of parameters by 18.1% and reduces the amount of computation by 15.5%. The improved model outperforms existing detection models in terms of model performance and accuracy, and is able to complete the recognition task in laboratory scenarios. The LUB-YOLO model is specifically designed for recognizing unsafe behaviors and is more suitable for deployment in laboratory scenarios with limited computational resources. This work provides theoretical insights for enhancing safety monitoring and reducing accidents.

人的不安全行为是造成实验室事故的主要原因；然而，现有的实验室监测方法在面对不安全行为时表现不佳。如何在不安全行为发生时准确、实时地预防被认为是一个关键挑战。本文提出了一种基于改进的YOLOv8s的实验室不安全行为检测- yolo模型（LUB-YOLO），解决了检测性能低、不安全行为难以识别的问题。该模型将Ghost Conv模块与HGNet相结合，优化了模型的计算复杂度，提高了检测性能；引入AIFI多头注意机制，增强了特征表示和结合上下文特征的能力；将DyHead的动态注意机制模块应用于检测头部，获取目标的尺度信息。实验结果表明，与原始网络相比，LUB-YOLO模型的[email protected]改进了2.1%，参数数量减少了18.1%，计算量减少了15.5%。改进后的模型在模型性能和准确率上都优于现有的检测模型，能够完成实验室场景下的识别任务。LUB-YOLO模型是专门为识别不安全行为而设计的，更适合在计算资源有限的实验室场景中部署。这项工作为加强安全监测和减少事故提供了理论见解。

{"title":"LUB-YOLO: A lightweight method for laboratory unsafe behavior detection based on improved YOLOv8s","authors":"Yulong Chen, Jiangming Kan, Chunjiang Yu","doi":"10.1016/j.measurement.2026.120623","DOIUrl":"10.1016/j.measurement.2026.120623","url":null,"abstract":"<div><div>Human unsafe behavior is the main cause of laboratory accidents; however, existing laboratory monitoring methods perform poorly in the face of unsafe behavior. How to prevent unsafe behavior accurately and in real-time when it occurs is considered a key challenge. In this paper, we propose a Laboratory Unsafe Behavior Detection-YOLO model (LUB-YOLO) based on the improved YOLOv8s, which solves the problems of low detection performance and difficulty in distinguishing unsafe behavior. The model combines the Ghost Conv module with HGNet to optimize the computational complexity of the model to improve the detection performance, incorporates the AIFI multi-head attention mechanism to enhance the feature representation and the ability to combine the contextual features, and the dynamic attention mechanism module of DyHead is applied to the detection head to obtain the scale information of the target. The experimental results show that compared with the original network, the LUB-YOLO model improves the [email protected] by 2.1%, reduces the number of parameters by 18.1% and reduces the amount of computation by 15.5%. The improved model outperforms existing detection models in terms of model performance and accuracy, and is able to complete the recognition task in laboratory scenarios. The LUB-YOLO model is specifically designed for recognizing unsafe behaviors and is more suitable for deployment in laboratory scenarios with limited computational resources. This work provides theoretical insights for enhancing safety monitoring and reducing accidents.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"269 ","pages":"Article 120623"},"PeriodicalIF":5.6,"publicationDate":"2026-02-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146147473","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

Performance enhancement of NRZ-based underwater optical wireless communication systems using square root detection under diverse water conditions 基于nrz的水下光无线通信系统在不同水条件下的平方根检测性能增强

IF 5.6 2区工程技术 Q1 ENGINEERING, MULTIDISCIPLINARY

Measurement

Pub Date : 2026-02-06 DOI: 10.1016/j.measurement.2026.120715

Ammar Armghan , Sultan S. Aldkeelalah , Slim Chaoui , Mehtab Singh , Somia A. Abd El-Mottaleb

In this paper, we present a detailed performance analysis of a non-return-to-zero (NRZ) based underwater optical wireless communication (UOWC) system under five standard water types: Pure Sea (PS), Clear Ocean (CO), Coastal Ocean (CS), Harbor I (HI), and Harbor II (HII). The analysis is carried out considering Q Factor, logarithmic bit error rate (log(BER)), received electrical power, and signal-to-noise ratio (SNR) as the key performance metrics. In the baseline system (without compensation), the maximum achievable ranges were limited to 52 m in PS, 30 m in CO, 17 m in CS, 8.6 m in HI, and 5.25 m in HII, with minimum log(BER) values in the range of –4.51 to –6.11 and Q Factors between 4.00 and 4.80 dB. To combat the non-linear response of the photodetector, we propose the deployment of a Square Root module (SRm) at the receiver. The incorporation of SRm demonstrates significant performance improvements, extending the maximum communication ranges to 150 m (PS), 65 m (CO), 32 m (CS), 14.5 m (HI), and 8.2 m (HII). This corresponds to nearly a threefold improvement in PS water and up to 60% enhancement in turbid harbor waters. Additionally, the system demonstrates improved detection quality, with Q Factors increasing to a range of 4.25–5.11 dB and log(BER) as low as –6.81 in PS and –6.65 in HII. Numerical simulations confirm that the proposed SRm-based detection technique effectively compensates for non-linear distortions and significantly enhances UOWC performance across diverse aquatic environments, making it a promising approach for medium to long range underwater optical links.

本文在纯海（PS）、清海（CO）、近海（CS）、海港I （HI）和海港II （HII）五种标准水域类型下，对基于非归零（NRZ）的水下光无线通信（UOWC）系统进行了详细的性能分析。分析考虑了Q因子、对数误码率（log(BER)）、接收电功率和信噪比（SNR）作为关键性能指标。在基线系统（无补偿）中，最大可实现范围限制为PS 52 m， CO 30 m, CS 17 m, HI 8.6 m和HII 5.25 m，最小对数（BER）值在-4.51至-6.11范围内，Q因子在4.00至4.80 dB之间。为了对抗光电探测器的非线性响应，我们建议在接收器上部署平方根模块（SRm）。SRm的结合显示了显著的性能改进，将最大通信范围扩展到150米（PS）， 65米（CO）， 32米（CS）， 14.5米（HI）和8.2米（HII）。这相当于PS水的近三倍改善和浑浊港口水域高达60%的改善。此外，该系统显示出更高的检测质量，Q因子增加到4.25-5.11 dB的范围，对数（BER）在PS和HII中低至-6.81和-6.65。数值模拟证实了基于srm的检测技术有效补偿了非线性失真，显著提高了不同水生环境下UOWC的性能，使其成为一种很有前途的中远距离水下光链路检测方法。

{"title":"Performance enhancement of NRZ-based underwater optical wireless communication systems using square root detection under diverse water conditions","authors":"Ammar Armghan , Sultan S. Aldkeelalah , Slim Chaoui , Mehtab Singh , Somia A. Abd El-Mottaleb","doi":"10.1016/j.measurement.2026.120715","DOIUrl":"10.1016/j.measurement.2026.120715","url":null,"abstract":"<div><div>In this paper, we present a detailed performance analysis of a non-return-to-zero (NRZ) based underwater optical wireless communication (UOWC) system under five standard water types: Pure Sea (PS), Clear Ocean (CO), Coastal Ocean (CS), Harbor I (HI), and Harbor II (HII). The analysis is carried out considering Q Factor, logarithmic bit error rate (log(BER)), received electrical power, and signal-to-noise ratio (SNR) as the key performance metrics. In the baseline system (without compensation), the maximum achievable ranges were limited to 52 m in PS, 30 m in CO, 17 m in CS, 8.6 m in HI, and 5.25 m in HII, with minimum log(BER) values in the range of –4.51 to –6.11 and Q Factors between 4.00 and 4.80 dB. To combat the non-linear response of the photodetector, we propose the deployment of a Square Root module (SRm) at the receiver. The incorporation of SRm demonstrates significant performance improvements, extending the maximum communication ranges to 150 m (PS), 65 m (CO), 32 m (CS), 14.5 m (HI), and 8.2 m (HII). This corresponds to nearly a threefold improvement in PS water and up to 60% enhancement in turbid harbor waters. Additionally, the system demonstrates improved detection quality, with Q Factors increasing to a range of 4.25–5.11 dB and log(BER) as low as –6.81 in PS and –6.65 in HII. Numerical simulations confirm that the proposed SRm-based detection technique effectively compensates for non-linear distortions and significantly enhances UOWC performance across diverse aquatic environments, making it a promising approach for medium to long range underwater optical links.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"269 ","pages":"Article 120715"},"PeriodicalIF":5.6,"publicationDate":"2026-02-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146147471","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

Quantifying graininess in metallic coatings using spatial and frequency-domain image descriptors: A comparison with visual perception 利用空间和频域图像描述子量化金属涂层的颗粒度：与视觉感知的比较

IF 5.6 2区工程技术 Q1 ENGINEERING, MULTIDISCIPLINARY

Measurement

Pub Date : 2026-02-06 DOI: 10.1016/j.measurement.2026.120706

Fatemeh Malekpour, Gorji Kandi Saeideh, Mohsen Mohseni

Graininess is a critical perceptual attribute of metallic coatings that strongly affects visual quality, yet its objective measurement remains challenging due to the lack of standardized and perceptually consistent metrics. This study proposes a camera-based image-processing framework for graininess evaluation, incorporating perceptually motivated preprocessing and multiscale analysis to achieve robust agreement with human visual perception while avoiding reliance on proprietary instrumental configurations. The framework is validated using a systematically designed dataset of automotive metallic coatings with controlled pigment characteristics and chromatic variations. Perceptual graininess is established through visual assessments by 32 observers and used as the reference for evaluation. Image-based texture descriptors from spatial and frequency domains are comparatively analyzed, with performance assessed using correlation- and error-based metrics and benchmarked against BYK-mac-i measurements. Results show that multiscale wavelet-based descriptors achieve the highest agreement with visual assessments (r = 0.94), outperforming both Fourier-based methods (r ≈ 0.92) and the industrial gonio-spectrophotometer (r ≈ 0.91). Overall, the proposed framework provides a transparent, computationally efficient, and perceptually consistent solution for reliable graininess quantification in industrial quality control.

颗粒度是金属涂层的关键感知属性，它强烈影响视觉质量，但由于缺乏标准化和感知一致的度量，其客观测量仍然具有挑战性。本研究提出了一种基于相机的图像处理框架，用于粒度评估，结合感知动机预处理和多尺度分析，以实现与人类视觉感知的强大一致性，同时避免依赖专有的仪器配置。使用系统设计的具有受控颜料特性和颜色变化的汽车金属涂层数据集验证了该框架。知觉粒度是通过32个观察者的视觉评估来建立的，并作为评价的参考。基于图像的空间和频域纹理描述符进行了比较分析，使用基于相关性和误差的度量来评估性能，并以BYK-mac-i测量值为基准。结果表明，基于多尺度小波的描述符与视觉评价的一致性最高（r = 0.94），优于基于傅里叶的方法（r≈0.92）和工业角形分光光度计（r≈0.91）。总体而言，所提出的框架为工业质量控制中的可靠粒度量化提供了一个透明、计算效率高、感知一致的解决方案。

{"title":"Quantifying graininess in metallic coatings using spatial and frequency-domain image descriptors: A comparison with visual perception","authors":"Fatemeh Malekpour, Gorji Kandi Saeideh, Mohsen Mohseni","doi":"10.1016/j.measurement.2026.120706","DOIUrl":"10.1016/j.measurement.2026.120706","url":null,"abstract":"<div><div>Graininess is a critical perceptual attribute of metallic coatings that strongly affects visual quality, yet its objective measurement remains challenging due to the lack of standardized and perceptually consistent metrics. This study proposes a camera-based image-processing framework for graininess evaluation, incorporating perceptually motivated preprocessing and multiscale analysis to achieve robust agreement with human visual perception while avoiding reliance on proprietary instrumental configurations. The framework is validated using a systematically designed dataset of automotive metallic coatings with controlled pigment characteristics and chromatic variations. Perceptual graininess is established through visual assessments by 32 observers and used as the reference for evaluation. Image-based texture descriptors from spatial and frequency domains are comparatively analyzed, with performance assessed using correlation- and error-based metrics and benchmarked against BYK-mac-i measurements. Results show that multiscale wavelet-based descriptors achieve the highest agreement with visual assessments (r = 0.94), outperforming both Fourier-based methods (r ≈ 0.92) and the industrial gonio-spectrophotometer (r ≈ 0.91). Overall, the proposed framework provides a transparent, computationally efficient, and perceptually consistent solution for reliable graininess quantification in industrial quality control.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"268 ","pages":"Article 120706"},"PeriodicalIF":5.6,"publicationDate":"2026-02-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146171995","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

DFA-Net: Dynamic multi-scale feature fusion and attention mechanism for surface defect detection in polysilicon production DFA-Net：多晶硅生产中表面缺陷检测的动态多尺度特征融合与关注机制

IF 5.6 2区工程技术 Q1 ENGINEERING, MULTIDISCIPLINARY

Measurement

Pub Date : 2026-02-06 DOI: 10.1016/j.measurement.2026.120754

Jiawen Sun , Wenzhong Yang , Yabo Yin , Jinhai Sa , Xinjun Pei , Fuyuan Wei , Danni Chen , Jianli Zhou

Accurate geometric measurement of steel surface defects is essential for quantitative quality assessment and process control in photovoltaic polysilicon production. Traditional methods primarily rely on fixed camera setups for qualitative inspection, with few studies exploring the use of unmanned aerial vehicles (UAVs). Existing deep learning-based detection approaches lack an intrinsic, calibrated metrology framework to establish traceable mapping from pixel-level outputs to physical dimensions. To address this challenge, this paper proposes a complete traceable visual metrology system. First, we established a calibrated UAV dynamic imaging platform. By establishing a known ground sampling distance (GSD = 1.2 mm/pixel), we constructed a deterministic mapping function from pixel coordinates to physical world coordinates. This function ensures each pixel corresponds to a clear and traceable physical scale. Second, we propose a specialized neural network architecture tailored for measurement tasks, serving as the core computational unit for quantitative measurement. Addressing the challenges of extreme defect size variations and complex background interference in industrial settings, we designed the measurement-driven DFA-Net. The Multi-scale Pyramid Pooling and Feature Processing (MSPP) module dynamically fuses cross-scale information to preserve complete defect features ranging from microcracks to macroscopic patches. The Omni-Dimensional Attention-Guided Selective Enhancement (ODESE) module improves spatial localization robustness under reflections, oil stains, and other interferences. The Wise-IoU v3 (WIoUv3) loss function ensures measurement consistency through dynamic gradient optimization based on defect difficulty and scale. Experimental results demonstrate that our method is effective in achieving traceable measurement outputs for physical parameters such as defect location and size range.

在光伏多晶硅生产中，钢表面缺陷的精确几何测量是定量质量评估和过程控制的关键。传统的方法主要依靠固定的摄像机设置进行定性检查，很少有研究探索无人驾驶飞行器（uav）的使用。现有的基于深度学习的检测方法缺乏一个内在的、校准的计量框架来建立从像素级输出到物理维度的可追溯映射。为了解决这一挑战，本文提出了一个完整的可追溯视觉计量系统。首先，建立了标定后的无人机动态成像平台。通过建立已知的地面采样距离（GSD = 1.2 mm/pixel），构建了从像素坐标到物理世界坐标的确定性映射函数。该功能确保每个像素对应于一个清晰和可追溯的物理尺度。其次，我们提出了一个专门针对测量任务的神经网络架构，作为定量测量的核心计算单元。为了解决工业环境中极端缺陷尺寸变化和复杂背景干扰的挑战，我们设计了测量驱动的DFA-Net。多尺度金字塔池化和特征处理（MSPP）模块动态融合跨尺度信息，以保留从微裂纹到宏观斑块的完整缺陷特征。全维注意引导选择性增强（ODESE）模块提高了在反射、油渍和其他干扰下的空间定位鲁棒性。WIoUv3 （Wise-IoU v3）损失函数通过基于缺陷难度和规模的动态梯度优化，保证测量一致性。实验结果表明，该方法可以有效地实现缺陷位置和尺寸范围等物理参数的可追溯测量输出。

{"title":"DFA-Net: Dynamic multi-scale feature fusion and attention mechanism for surface defect detection in polysilicon production","authors":"Jiawen Sun , Wenzhong Yang , Yabo Yin , Jinhai Sa , Xinjun Pei , Fuyuan Wei , Danni Chen , Jianli Zhou","doi":"10.1016/j.measurement.2026.120754","DOIUrl":"10.1016/j.measurement.2026.120754","url":null,"abstract":"<div><div>Accurate geometric measurement of steel surface defects is essential for quantitative quality assessment and process control in photovoltaic polysilicon production. Traditional methods primarily rely on fixed camera setups for qualitative inspection, with few studies exploring the use of unmanned aerial vehicles (UAVs). Existing deep learning-based detection approaches lack an intrinsic, calibrated metrology framework to establish traceable mapping from pixel-level outputs to physical dimensions. To address this challenge, this paper proposes a complete traceable visual metrology system. First, we established a calibrated UAV dynamic imaging platform. By establishing a known ground sampling distance (GSD = 1.2 mm/pixel), we constructed a deterministic mapping function from pixel coordinates to physical world coordinates. This function ensures each pixel corresponds to a clear and traceable physical scale. Second, we propose a specialized neural network architecture tailored for measurement tasks, serving as the core computational unit for quantitative measurement. Addressing the challenges of extreme defect size variations and complex background interference in industrial settings, we designed the measurement-driven DFA-Net. The Multi-scale Pyramid Pooling and Feature Processing (MSPP) module dynamically fuses cross-scale information to preserve complete defect features ranging from microcracks to macroscopic patches. The Omni-Dimensional Attention-Guided Selective Enhancement (ODESE) module improves spatial localization robustness under reflections, oil stains, and other interferences. The Wise-IoU v3 (WIoUv3) loss function ensures measurement consistency through dynamic gradient optimization based on defect difficulty and scale. Experimental results demonstrate that our method is effective in achieving traceable measurement outputs for physical parameters such as defect location and size range.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"269 ","pages":"Article 120754"},"PeriodicalIF":5.6,"publicationDate":"2026-02-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146147314","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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