Nowadays, various multimodal fusion-based 3-D object detectors appear to provide a potential opportunity to solve the failure cases in single-modality methods. However, current fusion approaches still face some unfavorable factors, e.g., poor illumination driving conditions and crowded traffic scenarios, which will cause unsatisfying image quality and objects’ occlusion. To this end, we present a multimodal fusion network FSFNet consisting of local graph-aware point cloud backbone (LGB), foreground score-aware fusion network (FSFN), and the proposals’ refining loss (PRL) for the 3-D object detection task in this article. Concretely, the directed graph is built to generate edgewise features for each point, and the point features are supplemented with graph information in LGB. To alleviate the defect of undesirable image quality features caused by poor illumination condition, FSFN is designed to produce an adaptive multimodal feature by taking pointwise foreground scores into consideration. Hence, levelwise point features with high confidence are fully used, and the imperfect image information is suppressed in fusion stage. We further introduce PRL to reduce the false positive and false negative cases in crowded scenes by optimizing the location and scores of predicted 3-D bounding boxes. Extensive experiments conducted on the KITTI benchmark demonstrate that FSFNet owns its superiority over state-of-the-art networks. Moreover, FSFN is verified to be robust against the image inputs under poor illumination conditions.

Measurement of temperature errors of fiber optic gyro (FOG) scale factor is important in FOG technology. In traditional methods, the measurement errors of FOG scale factors are greatly affected by the turntable’s orientation control errors. This article proposes a systematic method to measure the positive and negative FOG scale factor errors using a low-precision turntable. The relationship between the inertial navigation errors and the complete inertial measurement unit (IMU) errors is revealed. A simplified tracking algorithm is devised to obtain the change rates of the velocity errors with lower computation. A measurement procedure is designed, with three groups of $\pi $ and $- \pi $ rotations around the east axis of the turntable at each temperature. Observation equations are derived to estimate the positive and negative FOG scale factor errors. Simulation and experimental results show that a low-precision turntable can accurately measure the FOG scale factor errors at each temperature. Furthermore, temperature errors of accelerometer biases, scale factors, and nonlinear scale factors can also be measured at the same time using the rotation sequences and formulas in this article without additional rotations.

Wi-Fi signals-based passive gesture tracking is making quick progress in promoting human–computer interaction. The state-of-the-arts have achieved promising cm-level tracking accuracies via the Fresnel ellipses, while they depend heavily on the transceiver positions and the gesture initial positions. This problem is known as the position-dependence problem, which hinders the actual deployment of sensing applications. In this article, we propose EasyTrack, which aims at position-independency passive gesture tracking based on Wi-Fi channel state information (CSI). EasyTrack develops a novel incremental motion tracking model to correlate the length variations and the angle information of the reflection paths with the gesture traces. This model eliminates the dependence on the prior knowledge of transceiver positions and the gesture initial positions, by shifting the sensing observation from the traditional transceiver view to the antenna array-oriented view. We propose a series of tracking refinement approaches, including trace segmentation, tracking smoothing, and motion detection to improve the tracking accuracy. We prototype EasyTrack using off-the-shelf Wi-Fi radios and extensive experiments attest EasyTrack can achieve tracking accuracy of 0.9 cm under various users and environment conditions.

Pedestrian dead-reckoning (PDR) is a potential indoor localization technology that utilizes the inertial measurement unit (IMU) to estimate location. However, one of its most significant drawbacks is the accumulation measurement error. This article proposes a PDR system integrated with visible light positioning (VLP) that boasts low accumulated error and can achieve real-time and accurate indoor positioning using both the IMU and the camera sensor on a smartphone. A multiframe fusion method for light-emitting diode (LED)-ID decoding is introduced in the encoding and decoding process of the system, achieving 98.5% decoding accuracy with a 20-bit-long LED-ID at the height of 2.1 m. This method also accommodates the variation in the shutter speeds of cameras and the heights of the LED. Meanwhile, absolute locations and step length could be further calibrated using a single LED, providing an average accuracy within 0.5 m over a 108-m walk.

The practical need for the accurate detection of human activities requires wearable flexible strain sensors that can detect subtle and large strains and recognize the bending direction of joints. However, it remains challenging to integrate these detection capabilities into a single strain sensor. Herein, we prepared flexible strain sensors with firmly attached locally warped graphene on the surface of anhydride-grafted styrene-butadiene-styrene triblock copolymer films using a swelling-self-assembly method. The unique localized warpage structure of graphene sheets gives the sensor excellent strain response performance and the ability to recognize the bending direction. The detection limit of the sensor to strain is independent of the sensitivity to strain, which allows it to detect ultralow (0.01%) and large strains (500%). Moreover, the sensor has a large linear response range (0%–94%) and good test stability (3000 cycles of strain). The sensor can be used to recognize the bending direction by generating opposite response signals to different bending directions. The excellent response behavior allows the sensor to accurately monitor the joint bending angle, the direction of the human body, and subtle tremors of the human skin, which makes it a potential application for wearable devices and smart prosthetics.

In practice, forging processes have many unknown dynamics and fast time-varying characteristics due to transient load variation. This often results in insufficient data samples. Usually, support vector regression (SVR) can accurately model small sample data due to its good sparsity. However, using sequential minimum optimization (SMO) algorithms increases the computational costs during the solution process, making it difficult to model fast time-varying systems. Aiming to address this problem, we developed an approach using online stochastic gradient descent (SGD) to improve SVR modeling efficiency. First, we used loss coefficients derived from the loss function to represent support vector loss. This effectively ensured sparsity in the modeling process. In this way, the SVR solving process was transformed into a loss coefficient calculation. This calculation was easy to achieve using SGD; thus, the solving process complexity was greatly reduced compared to SMO. On this basis, we developed an online incremental strategy to adapt the time-varying dynamics using online updating of step length, loss coefficients, and bias term. Additional analysis demonstrated the convergence of the proposed online modeling method. Furthermore, the modeling effect of this method is verified by using actual experiments with a 40-MN isothermal die forging press. Index Data-driven model, deformation force, die forging system, online model, support vector regression (SVR).

Ultraviolet (UV) photodetectors (PDs) have attracted significant attention for civil and military applications. Zinc oxide (ZnO) and nickel oxide (NiO) have been widely applied in UV-PDs due to their wide bandgap (3.2–3.8 eV), transparency, excellent optical and electrical properties, and good chemical stability. However, UV signals are generally weak; hence, UV-PDs with high optical gain are essential. In this work, high-performance solar-blind p-NiO/n-ZnO/p-Si heterojunction bipolar phototransistors (HBPTs) were fabricated. The fabricated HBPTs exhibited a high responsivity of $9.4\times 10^{{3}}$ A/W at a wavelength of 280 nm with ${V}_{\text {CE}}$ = −7 V, a high optical gain of $3.96\times 10^{{4}}$ , and a large detectivity of $3\times 10^{{13}}$ Jones. In addition, the UV/visible rejection ratio was as high as 880. These behaviors indicate that the prepared HBPTs are good solar-blind PDs and suitable for the detection of weak UV signals. However, for ${V}_{\text {CE}}$ value below −7 V, ( $\vert {V}_{\text {CE}}\vert >$ 7 V), the optical gain decreased due to the punchthrough effect. Furthermore, band diagrams showed that the photogenerated electrons were blocked by the potential barrier at the NiO/ZnO interface (base–emitter junction) due to a large conduction-band discontinuity ( $\Delta {E}_{C}{)}$ of 2.7 eV, which resulted in a large optical gain in the prepared HBPTs.

There are several complex situations in recognizing concealed objects from millimeter wave multiple-input multiple-output synthetic aperture radar (MIMO-SAR) security images, such as incomplete imaging of objects, partially occluded objects, and overlapping objects, which are detrimental to the accurate recognition of concealed objects. To solve these problems, a concealed object detection method based on a high-resolution feature recursive alignment fusion network (HR-FRAFnet) is proposed. The HR-FRAFnet can segment the object area from the grayscale image with complex human background and complete the recognition. The overall architecture of the HR-FRAFnet follows the encoder–decoder framework. Specifically, in the encoder stage, a deep parallel feature extraction network (DPFEN) connects the multiresolution feature maps in parallel and repeats multiscale feature fusion. This approach suppresses the background noise flowing and retains more recognizable target characteristics. Then, in the decoder stage, a feature recursive alignment fusion module (FRAFM) is designed to enhance the perception of object edges. The FRAFM effectively improves the segmentation accuracy of objects while decreasing the computational complexity of the network. Besides, we employ a combined loss function to alleviate the foreground-background imbalance problem in MIMO-SAR images. Homemade human security screening image datasets are used for evaluation. The experimental results show that the proposed method outperforms existing semantic segmentation methods in mean intersection over union (mIoU) and reduces the incidence of missed and error detection of targets.

In this article, a novel approach is proposed to detect the concentration of virus with different refractive indices in saliva using 2-D photonic crystal (PC) structures. The proposed $18\times17$ hexagonal PC structure consists of air holes in a silicon slab with silicon-on-insulator (SOI) techniques. The plane wave expansion (PWE) method is used to find the photonic band gap (PBG). The mid-gap wavelength of PBG increases linearly with a refractive index (RI) of infiltrated samples in air holes. The change in PBG size increases linearly by an average slope of 4.45 with respect to the change in RI concentration. The proposed sensor exhibits a maximum sensitivity of 1206.4 nm/RIU for 100% concentration of virus and minimum sensitivity of 950.4 nm/RIU for 0% concentration of virus with an ultra-fast average response time of 10.6 fs to test the samples. The present work can be useful for sensing variation in bio-molecular RI and other infection’s presence in various biological targets at an early stage.

Contemporary advanced driver assistance system (ADAS) features for semi-autonomous vehicles include braking assistance during collision avoidance. Although precollision detection typically relies on sensing systems to enable production vehicles to perceive oncoming road obstacles, the physiological state of the driver is not measured to predict emergency braking. On the other hand, previous driving simulation experiments have demonstrated the ability of regularized linear discriminant analysis (RLDA) to predict precollision braking using brain signals from multiple electroencephalogram (EEG) electrodes. In contrast, the current study used EEG data from these previous experiments to determine the quality of support vector machine (SVM) predictions as a first step toward realizing a brain–computer interface (BCI) for emergency braking. Power spectral density (PSD) features were extracted from the EEG of one electrode to train and evaluate an SVM. Through a novel data ablation analysis, the optimal number of PSD components was determined to optimize model classification quality measured by the area under the curve (AUC). A comparison of the proposed model to the previous RLDA and other machine learning (ML) methods indicated that the SVM had a superior AUC. Thus, the proposed model is a candidate for assisting ADASs with precollision detection. Moreover, since the proposed model only utilized one electrode, our study potentially contributes to the facilitation of BCIs for autonomous vehicles.

Ninety percent of the world’s cargo is transported by sea, and the fatigue of ship officers of the watch (OOWs) contributes significantly to maritime accidents. The fatigue detection of ship OOWs is more difficult than that of vehicle drivers due to an increase in the automation degree. In this study, research progress pertaining to fatigue detection in OOWs is comprehensively analyzed based on a comparison with that in vehicle drivers. Fatigue detection techniques for OOWs are organized based on input sources, which include the physiological/behavioral features of OOWs, vehicle/ship features, and their comprehensive features. Prerequisites for detecting fatigue in OOWs are summarized. Subsequently, various input features applicable and existing applications to the fatigue detection of OOWs are proposed, and their limitations are analyzed. The results show that the reliability of the acquired feature data is insufficient for detecting fatigue in OOWs, as well as a nonnegligible invasive effect on OOWs. Hence, low-invasive physiological information pertaining to the OOWs, behavior videos, and multisource feature data of ship characteristics should be used as inputs in future studies to realize quantitative, accurate, and real-time fatigue detections in OOWs on actual ships.

In this article, we propose a method for performing the simultaneous localization and mapping (SLAM) in indoor environments using only multiple-input and multiple-output (MIMO) frequency-modulated continuous wave (FMCW) radar. The SLAM is a technology that maps the surrounding environment and position of a platform simultaneously. The following are the steps of the overall the SLAM implementation of the proposed method. First, the ego-velocity of the platform is estimated using the relative velocity with respect to stationary objects. In this case, a random sample consensus (RANSAC) algorithm is used in the velocity–angle map generated from the detection results to identify stationary objects. Second, the rotation angle of the platform is estimated using the linear component extracted from the walls in the ${xy}$ plane. Then, we determine the ego-motion of the platform using the estimated ego-velocity and rotation angle. Finally, we map the position of the platform and the indoor environment simultaneously in the absolute coordinate system. We validate the mapping result generated using our proposed method by comparing it with the ground truth-based mapping result.

In this article, a constrained weighted least-squares (CWLS) problem with two quadratic constraints for hybrid angle of arrival (AOA), time difference of arrival (TDOA), and received signal strength (RSS) localization is formulated. The Lagrange multiplier method cannot be efficiently used in this situation, so we provide four methods to solve the CWLS problem: 1) weighted least-squares (WLS); 2) iterative CWLS (ICWLS); 3) Lagrange multiplier method combined with Newton’s method (LMM+NM); and 4) semidefinite programming (SDP). Furthermore, to obtain a good balance between accuracy and computational burden, we develop a fixed point iteration (FPI)-based algorithm that is derived from an approximate maximum likelihood estimator (MLE). Simulation results show that the accuracy of the proposed FPI method is better than the other four methods; besides, its computational burden is one in ten of the SDP algorithm.

Good manufacturing practice for medicinal products is laid down in several guidelines and Directives of the European Commission. Those regulations imply, among other aspects, that medicinal product manufacturers have to ensure that the final products are fit for their intended use and do not place patients at risk due to the inadequate safety, quality, or efficacy. For the case of manufacturing of pharmaceutical blisters, the attainment of this quality objective often leads to the resourcing of qualified personnel for final visual verification of the blister pack content. The need for inline content verification of pharmaceutical blisters asks therefore for sensors that provide fast, noncontact, and accurate chemical information of each individual blister content. Here, we report on a quantum cascade laser (QCL)-based blister-verification sensor. The verification principle is substance chemical identification by means of backscattering mid-infrared (IR) spectroscopy. The light source is a palm-size wavelength-tunable mid-IR QCL with $\sim $ 1-kHz tuning speed. The blister content verification uses machine vision to obtain the required position information for each individual content and fast spatial scanning facilitated by a two-axis galvanometer scanner. Diffuse reflectance mid-IR spectra are acquired at each location, and their classification is conducted instantaneously. Different classifier approaches are evaluated and discussed including machine learning and standard cross correlation to Fourier-transform-IR (FTIR) data. Altogether, this sensor is capable of scanning a standard 12-pill blister pack in $\sim $ 0.3 s, whereas this scanning time is essentially related to the desired classification accuracy, but not to the spectral resolution, which is fixed. Using machine learning classification, 100% identification accuracy is demonstrated for 13 different medication types (i.e., with different chemical nature), whereas only 97.4% identification accuracy is achieved by standard cross correlation to FTIR data. The used pills have all similar size, shape, and color, so that classification by visual inspection is barely possible.

Ship detection is of great significance in the interpretation of synthetic aperture radar (SAR) images. However, SAR generates inherent speckle noise when producing images, which poses many challenges for ship detection tasks. One major issue during detection is low accuracy caused by noise interference near the ship. To address this issue, this study aims to design a deep dense attention detection network for improving the accuracy of ship target detection in SAR. The proposed algorithm primarily uses a multilayer deep dense network to preliminarily extract ship image features and subsequently introduces an attention network to further enhance these features. Finally, an anchor point mechanism is utilized to perform ship positioning regression estimation. Experimental results on public SAR ship datasets, including SAR ship detection dataset (SSDD) and SAR-Ship-Dataset, demonstrate that the proposed algorithm performs well in terms of speed and accuracy and has better robustness and real-time performance compared to similar detection algorithms.

Optical fiber inclinometer technology has been gradually used in various projects. However, the error analysis of this method is decisive for its subsequent development and application. In this study, through three tests, the monitoring length, layout plane, and boundary conditions that may lead to measurement errors are studied in detail and the contribution of different factors to measurement errors is explored. The research results reveal that the measurement errors caused by monitoring length and layout plane have an empirical function relationship with the monitoring length ${x}$ and the angle $\theta $ between the layout plane and the inclined direction. Any changes in the boundary conditions will result in an overall deviation in the calculated deflection curve. After analyzing the characteristics of each error, a comprehensive error equation is proposed. This equation has a guiding significance for error correction and technical improvement in optical fiber inclinometer.

Current-based fault diagnosis has become a promising solution for electromechanical systems due to the low cost and easy access. However, most of the existing studies require additional signal processing or diagnostic expertise to preprocess the signals due to the effects of the fundamental component and electrical noise. It is challenging to extract effective fault-related features from the raw current signals. To this end, this article proposes a differential-augmented current feature learning network named CurrentNet for drive system fault diagnosis, which is an end-to-end model. First, a differential-augmented strategy based on the raw current signal is introduced to generate complementary current representations. Furthermore, a multi-information interaction module (MIIM) is designed to adaptively capture complementary shared information between the original and enhanced signals through a parallel mechanism. Also, efficient channel attention is used to reduce the complexity of the model. Our proposed method is experimentally evaluated on two datasets and presents obvious superiority over existing fault diagnosis methods.

Worm-like robots that mimic the peristaltic locomotion of earthworms have high robustness to complex environments. These robots’ movements are driven by the deformation of the body, but this compliance in the body brings challenges for tracking and control. This work compares three sensing methods to track a worm-like robot’s displacement: the actuator forward modeling method (AFMM), the stretch and pressure sensor method (SPSM), and the inertial measurement unit method (IMUM). Each of these methods is compared against a true displacement determined by vision tracking. Based on experimental results, SPSM yields the lowest average error (underestimating the true value by 17% on average), AFMM is slightly higher (20% underestimation), and the IMU result has a comparatively large average error (77% overestimation). AFMM failed to track the robot’s backward slip, while both SPSM and IMUM showed the ability of slip detection.

Presence of excess urea in milk above the maximum permissible limit can result in serious health issues, including kidney failure. While several tests exist for measuring urea concentration in milk, such tests typically require significant preprocessing of samples in a laboratory setting and result in a long turnaround time and increased cost of testing due to the use of expensive equipment operated by trained analysts. In this work, we present a point-of-use sensor for the quantification of urea concentration in milk samples, with minimal sample preprocessing and seamless readout. The assay involves the addition of p-dimethylaminobenzaldehyde (p-DMAB) to the supernatant obtained after the precipitation of milk solids, to form a yellow-green color chromogen. The intensity of the color is proportional to the amount of chromogen produced and is probed using a low-cost optical phase sensitive detection (PSD) system assembled using high-performance albeit inexpensive consumer-grade components. The data acquired by the sensor is streamed wirelessly to a smartphone for further analysis. We present preliminary results obtained using milk samples spiked with urea, with concentrations ranging from 0.1 mg/mL (safe level) to 0.7 mg/mL (maximum permissible limit), with a limit of detection (LOD) of 0.19 mg/mL.

Wearable sensors and the Internet of Things (IoT) will be two buzzwords that will be heard commonly in the coming decades. The combination of these two technologies soon will create a great revolution in applications that require motion recognition, such as health care, sports, and entertainment. The development of technology has made wearable sensors one of the most basic tools for human motion analysis. We believe that IoT is the most powerful complement to the use of wearable sensors in the analysis of human body motion. Using wearable IoT, all necessary human data will be collected and delivered via the Internet to the experts who can make accurate decisions about the type of activity, falling situations, freezing of gait (fog), and so on. In this article, the human motion analysis is presented in a chart and is divided into two parts: movement measurement and movement classification. However, this article focuses on movement classification that includes three subsections, gait analysis (GA), gesture recognition (GR), and human activity recognition (HAR), and is closely related to human motion recognition. In this article, our goal is to first acquaint the reader with the important steps required to classify the movement of the human body by wearable sensors and then by using tables to determine the most used algorithms and methods for each step. After briefly reviewing IoT concepts, directions for further research will be provided.