Next Issue
Volume 21, October-2
Previous Issue
Volume 21, September-2
sensors-logo

Journal Browser

Journal Browser

Sensors, Volume 21, Issue 19 (October-1 2021) – 373 articles

Cover Story (view full-size image): Flexible and biodegradable sensors are advantageous for their versatility in a range of areas from agriculture to smart packaging. This work characterizes the performance of capacitive relative humidity sensors printed on different biodegradable substrates. The behavior of polylactide (PLA), glossy paper, and potato starch as a sensing layer is compared to that of nonbiodegradable polyethylene terephthalate (PET). Capacitance across the inkjet-printed electrodes is measured in controlled environmental conditions ranging from 15 to 90% relative humidity. The sensitivity, response time, hysteresis, and thermal drift are tested and compared for each sensor. This work shows potential for the use of biodegradable sensors as renewable materials become more widespread. View this paper
  • Issues are regarded as officially published after their release is announced to the table of contents alert mailing list.
  • You may sign up for e-mail alerts to receive table of contents of newly released issues.
  • PDF is the official format for papers published in both, html and pdf forms. To view the papers in pdf format, click on the "PDF Full-text" link, and use the free Adobe Readerexternal link to open them.
Order results
Result details
Select all
Export citation of selected articles as:
Article
Multi-Scale Capsule Attention Network and Joint Distributed Optimal Transport for Bearing Fault Diagnosis under Different Working Loads
Sensors 2021, 21(19), 6696; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196696 - 08 Oct 2021
Viewed by 394
Abstract
In recent years, intelligent fault diagnosis methods based on deep learning have developed rapidly. However, most of the existing work performs well under the assumption that training and testing samples are collected from the same distribution, and the performance drops sharply when the [...] Read more.
In recent years, intelligent fault diagnosis methods based on deep learning have developed rapidly. However, most of the existing work performs well under the assumption that training and testing samples are collected from the same distribution, and the performance drops sharply when the data distribution changes. For rolling bearings, the data distribution will change when the load and speed change. In this article, to improve fault diagnosis accuracy and anti-noise ability under different working loads, a transfer learning method based on multi-scale capsule attention network and joint distributed optimal transport (MSCAN-JDOT) is proposed for bearing fault diagnosis under different loads. Because multi-scale capsule attention networks can improve feature expression ability and anti-noise performance, the fault data can be better expressed. Using the domain adaptation ability of joint distribution optimal transport, the feature distribution of fault data under different loads is aligned, and domain-invariant features are learned. Through experiments that investigate bearings fault diagnosis under different loads, the effectiveness of MSCAN-JDOT is verified; the fault diagnosis accuracy is higher than that of other methods. In addition, fault diagnosis experiment is carried out in different noise environments to demonstrate MSCAN-JDOT, which achieves a better anti-noise ability than other transfer learning methods. Full article
(This article belongs to the Section Intelligent Sensors)
Show Figures

Figure 1

Article
Validation of Low-Cost Impedance Analyzer via Nitrate Detection
Sensors 2021, 21(19), 6695; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196695 - 08 Oct 2021
Viewed by 348
Abstract
Impedance spectroscopy is a widely used electrochemical technique with a wide variety of applications. Many of these applications benefit from the additional accessibility provided by low-cost impedance devices. With this in mind, a low-cost impedance device was designed for a high performance-to-cost ratio. [...] Read more.
Impedance spectroscopy is a widely used electrochemical technique with a wide variety of applications. Many of these applications benefit from the additional accessibility provided by low-cost impedance devices. With this in mind, a low-cost impedance device was designed for a high performance-to-cost ratio. The performance of this analyzer was validated against a high-performance DropSens µStat-i 400s potentiostat by performing an application-based experiment. Nitrate detection provides a relevant experiment because of the importance of maintaining precise nitrate concentrations to mitigate the impact of nitrate fluctuations on the environment. Dissolved nitrate samples of different concentrations, in the range 3–1000 mg/L, were confirmed colorimetrically and measured with both instruments. A calibration curve of the real impedance matched a sigmoidal transfer, with a linear region for concentrations below 10 mg/L. The device under investigation exhibited an average magnitude error of 1.28% and an average phase error of 0.96 relative to the high-performance standard, which validates the performance of the low-cost device. A cost analysis is presented that highlights some of the complexities of cost comparisons. Full article
(This article belongs to the Section Electronic Sensors)
Show Figures

Figure 1

Article
A Novel Capacitive Measurement Device for Longitudinal Monitoring of Bone Fracture Healing
Sensors 2021, 21(19), 6694; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196694 - 08 Oct 2021
Viewed by 401
Abstract
The healing process of surgically-stabilised long bone fractures depends on two main factors: (a) the assessment of implant stability, and (b) the knowledge of bone callus stiffness. Currently, X-rays are the main diagnostic tool used for the assessment of bone fractures. However, they [...] Read more.
The healing process of surgically-stabilised long bone fractures depends on two main factors: (a) the assessment of implant stability, and (b) the knowledge of bone callus stiffness. Currently, X-rays are the main diagnostic tool used for the assessment of bone fractures. However, they are considered unsafe, and the interpretation of the clinical results is highly subjective, depending on the clinician’s experience. Hence, there is the need for objective, non-invasive and repeatable methods to allow a longitudinal assessment of implant stability and bone callus stiffness. In this work, we propose a compact and scalable system, based on capacitive sensor technology, able to measure, quantitatively, the relative pins displacements in bone fractures treated with external fixators. The measurement device proved to be easily integrable with the external fixator pins. Smart arrangements of the sensor units were exploited to discriminate relative movements of the external pins in the 3D space with a resolution of 0.5 mm and 0.5°. The proposed capacitive technology was able to detect all of the expected movements of the external pins in the 3D space, providing information on implant stability and bone callus stiffness. Full article
(This article belongs to the Special Issue Robotic Sensing for Biomedical Applications)
Show Figures

Graphical abstract

Article
Internet of Things (IoT)-Enhanced Applied Behavior Analysis (ABA) for Special Education Needs
Sensors 2021, 21(19), 6693; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196693 - 08 Oct 2021
Viewed by 497
Abstract
Applied behavior analysis (ABA) has become a popular behavioral therapy in the special education needs (SEN) community. ABA is used to manage SEN students’ behaviors by solving problems in socially important settings, and puts emphasis on having precise measurements on physical and observable [...] Read more.
Applied behavior analysis (ABA) has become a popular behavioral therapy in the special education needs (SEN) community. ABA is used to manage SEN students’ behaviors by solving problems in socially important settings, and puts emphasis on having precise measurements on physical and observable events. In this work, we present how Internet of Things (IoT) technologies can be applied to enhance ABA therapy in normal SEN classroom settings. We measured (1) learning performance data, (2) learners’ physiological data, and (3) learning environment sensors’ data. Upon preliminary analysis, we have found that learners’ physiological data is highly diverse, while learner performance seems to be related to learners’ electrodermal activity. Our preliminary findings suggest the possibility of enhancing ABA for SEN with IoT technologies. Full article
(This article belongs to the Section Biosensors)
Show Figures

Figure 1

Article
A Portable Smart Fitness Suite for Real-Time Exercise Monitoring and Posture Correction
Sensors 2021, 21(19), 6692; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196692 - 08 Oct 2021
Viewed by 382
Abstract
Fitness and sport have drawn significant attention in wearable and persuasive computing. Physical activities are worthwhile for health, well-being, improved fitness levels, lower mental pressure and tension levels. Nonetheless, during high-power and commanding workouts, there is a high likelihood that physical fitness is [...] Read more.
Fitness and sport have drawn significant attention in wearable and persuasive computing. Physical activities are worthwhile for health, well-being, improved fitness levels, lower mental pressure and tension levels. Nonetheless, during high-power and commanding workouts, there is a high likelihood that physical fitness is seriously influenced. Jarring motions and improper posture during workouts can lead to temporary or permanent disability. With the advent of technological advances, activity acknowledgment dependent on wearable sensors has pulled in countless studies. Still, a fully portable smart fitness suite is not industrialized, which is the central need of today’s time, especially in the Covid-19 pandemic. Considering the effectiveness of this issue, we proposed a fully portable smart fitness suite for the household to carry on their routine exercises without any physical gym trainer and gym environment. The proposed system considers two exercises, i.e., T-bar and bicep curl with the assistance of the virtual real-time android application, acting as a gym trainer overall. The proposed fitness suite is embedded with a gyroscope and EMG sensory modules for performing the above two exercises. It provided alerts on unhealthy, wrong posture movements over an android app and is guided to the best possible posture based on sensor values. The KNN classification model is used for prediction and guidance for the user while performing a particular exercise with the help of an android application-based virtual gym trainer through a text-to-speech module. The proposed system attained 89% accuracy, which is quite effective with portability and a virtually assisted gym trainer feature. Full article
(This article belongs to the Special Issue Inertial Motion Capture and Sensing Technologies)
Show Figures

Figure 1

Article
Automatic Prediction of Ischemia-Reperfusion Injury of Small Intestine Using Convolutional Neural Networks: A Pilot Study
Sensors 2021, 21(19), 6691; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196691 - 08 Oct 2021
Viewed by 284
Abstract
Acute intestinal ischemia is a life-threatening condition. The current gold standard, with evaluation based on visual and tactile sensation, has low specificity. In this study, we explore the feasibility of using machine learning models on images of the intestine, to assess small intestinal [...] Read more.
Acute intestinal ischemia is a life-threatening condition. The current gold standard, with evaluation based on visual and tactile sensation, has low specificity. In this study, we explore the feasibility of using machine learning models on images of the intestine, to assess small intestinal viability. A digital microscope was used to acquire images of the jejunum in 10 pigs. Ischemic segments were created by local clamping (approximately 30 cm in width) of small arteries and veins in the mesentery and reperfusion was initiated by releasing the clamps. A series of images were acquired once an hour on the surface of each of the segments. The convolutional neural network (CNN) has previously been used to classify medical images, while knowledge is lacking whether CNNs have potential to classify ischemia-reperfusion injury on the small intestine. We compared how different deep learning models perform for this task. Moreover, the Shapley additive explanations (SHAP) method within explainable artificial intelligence (AI) was used to identify features that the model utilizes as important in classification of different ischemic injury degrees. To be able to assess to what extent we can trust our deep learning model decisions is critical in a clinical setting. A probabilistic model Bayesian CNN was implemented to estimate the model uncertainty which provides a confidence measure of our model decisions. Full article
(This article belongs to the Special Issue Innovative Sensors and Biosensors for In Vitro/In Vivo Diagnostics)
Show Figures

Figure 1

Review
Modern Types of Axicons: New Functions and Applications
Sensors 2021, 21(19), 6690; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196690 - 08 Oct 2021
Viewed by 309
Abstract
Axicon is a versatile optical element for forming a zero-order Bessel beam, including high-power laser radiation schemes. Nevertheless, it has drawbacks such as the produced beam’s parameters being dependent on a particular element, the output beam’s intensity distribution being dependent on the quality [...] Read more.
Axicon is a versatile optical element for forming a zero-order Bessel beam, including high-power laser radiation schemes. Nevertheless, it has drawbacks such as the produced beam’s parameters being dependent on a particular element, the output beam’s intensity distribution being dependent on the quality of element manufacturing, and uneven axial intensity distribution. To address these issues, extensive research has been undertaken to develop nondiffracting beams using a variety of advanced techniques. We looked at four different and special approaches for creating nondiffracting beams in this article. Diffractive axicons, meta-axicons-flat optics, spatial light modulators, and photonic integrated circuit-based axicons are among these approaches. Lately, there has been noteworthy curiosity in reducing the thickness and weight of axicons by exploiting diffraction. Meta-axicons, which are ultrathin flat optical elements made up of metasurfaces built up of arrays of subwavelength optical antennas, are one way to address such needs. In addition, when compared to their traditional refractive and diffractive equivalents, meta-axicons have a number of distinguishing advantages, including aberration correction, active tunability, and semi-transparency. This paper is not intended to be a critique of any method. We have outlined the most recent advancements in this field and let readers determine which approach best meets their needs based on the ease of fabrication and utilization. Moreover, one section is devoted to applications of axicons utilized as sensors of optical properties of devices and elements as well as singular beams states and wavefront features. Full article
(This article belongs to the Special Issue Recent Developments of Integrated Photonic Sensors)
Show Figures

Figure 1

Article
Coresets for the Average Case Error for Finite Query Sets
Sensors 2021, 21(19), 6689; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196689 - 08 Oct 2021
Viewed by 308
Abstract
Coreset is usually a small weighted subset of an input set of items, that provably approximates their loss function for a given set of queries (models, classifiers, hypothesis). That is, the maximum (worst-case) error over all queries is bounded. To obtain smaller coresets, [...] Read more.
Coreset is usually a small weighted subset of an input set of items, that provably approximates their loss function for a given set of queries (models, classifiers, hypothesis). That is, the maximum (worst-case) error over all queries is bounded. To obtain smaller coresets, we suggest a natural relaxation: coresets whose average error over the given set of queries is bounded. We provide both deterministic and randomized (generic) algorithms for computing such a coreset for any finite set of queries. Unlike most corresponding coresets for the worst-case error, the size of the coreset in this work is independent of both the input size and its Vapnik–Chervonenkis (VC) dimension. The main technique is to reduce the average-case coreset into the vector summarization problem, where the goal is to compute a weighted subset of the n input vectors which approximates their sum. We then suggest the first algorithm for computing this weighted subset in time that is linear in the input size, for n1/ε, where ε is the approximation error, improving, e.g., both [ICML’17] and applications for principal component analysis (PCA) [NIPS’16]. Experimental results show significant and consistent improvement also in practice. Open source code is provided. Full article
(This article belongs to the Special Issue Sensor Data Summarization: Theory, Applications, and Systems)
Show Figures

Figure 1

Article
Building an IoT Platform Based on Service Containerisation
Sensors 2021, 21(19), 6688; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196688 - 08 Oct 2021
Viewed by 318
Abstract
IoT platforms have become quite complex from a technical viewpoint, becoming the cornerstone for information sharing, storing, and indexing given the unprecedented scale of smart services being available by massive deployments of a large set of data-enabled devices. These platforms rely on structured [...] Read more.
IoT platforms have become quite complex from a technical viewpoint, becoming the cornerstone for information sharing, storing, and indexing given the unprecedented scale of smart services being available by massive deployments of a large set of data-enabled devices. These platforms rely on structured formats that exploit standard technologies to deal with the gathered data, thus creating the need for carefully designed customised systems that can handle thousands of heterogeneous data sensors/actuators, multiple processing frameworks, and storage solutions. We present the SCoT2.0 platform, a generic-purpose IoT Platform that can acquire, process, and visualise data using methods adequate for both real-time processing and long-term Machine Learning (ML)-based analysis. Our goal is to develop a large-scale system that can be applied to multiple real-world scenarios and is potentially deployable on private clouds for multiple verticals. Our approach relies on extensive service containerisation, and we present the different design choices, technical challenges, and solutions found while building our own IoT platform. We validate this platform supporting two very distinct IoT projects (750 physical devices), and we analyse scaling issues within the platform components. Full article
(This article belongs to the Section Communications)
Show Figures

Figure 1

Article
Tool Condition Monitoring of the Cutting Capability of a Turning Tool Based on Thermography
Sensors 2021, 21(19), 6687; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196687 - 08 Oct 2021
Viewed by 342
Abstract
In turning, the wear control of a cutting tool benefits product quality enhancement, tool-related costs‘ optimisation, and assists in avoiding undesired events. In small series and individual production, the machine operator is the one who determines when to change a cutting tool, based [...] Read more.
In turning, the wear control of a cutting tool benefits product quality enhancement, tool-related costs‘ optimisation, and assists in avoiding undesired events. In small series and individual production, the machine operator is the one who determines when to change a cutting tool, based upon their experience. Bad decisions can often lead to greater costs, production downtime, and scrap. In this paper, a Tool Condition Monitoring (TCM) system is presented that automatically classifies tool wear of turning tools into four classes (no, low, medium, high wear). A cutting tool was monitored with infrared (IR) camera immediately after the cut and in the following 60 s. The Convolutional Neural Network Inception V3 was used to analyse and classify the thermographic images, which were divided into different groups depending on the time of acquisition. Based on classification result, one gets information about the cutting capability of the tool for further machining. The proposed model, combining Infrared Thermography, Computer Vision, and Deep Learning, proved to be a suitable method with results of more than 96% accuracy. The most appropriate time of image acquisition is 6–12 s after the cut is finished. While existing temperature based TCM systems focus on measuring a cutting tool absolute temperature, the proposed system analyses a temperature distribution (relative temperatures) on the whole image based on image features. Full article
Show Figures

Figure 1

Article
Implementation of Facility Management for Port Infrastructure through the Use of UAVs, Photogrammetry and BIM
Sensors 2021, 21(19), 6686; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196686 - 08 Oct 2021
Viewed by 301
Abstract
The maintenance of port infrastructures presents difficulties due to their location: an aggressive environment or the variability of the waves can cause progressive deterioration. Maritime conditions make inspections difficult and, added to the lack of use of efficient tools for the management of [...] Read more.
The maintenance of port infrastructures presents difficulties due to their location: an aggressive environment or the variability of the waves can cause progressive deterioration. Maritime conditions make inspections difficult and, added to the lack of use of efficient tools for the management of assets, planning maintenance, important to ensure operability throughout the life cycle of port infrastructures, is generally not a priority. In view of these challenges, this research proposes a methodology for the creation of a port infrastructure asset management tool, generated based on the Design Science Research Method (DSRM), in line with Building Information Modeling (BIM) and digitization trends in the infrastructure sector. The proposal provides workflows and recommendations for the survey of port infrastructures from UAVs, the reconstruction of digital models by photogrammetry (due to scarce technical documentation), and the reconstruction of BIM models. Along with this, the bidirectional linking of traditional asset management spreadsheets with BIM models is proposed, by visual programming, allowing easy visualization of the status and maintenance requirements. This methodology was applied to a port infrastructure, where the methodology demonstrated the correct functionality of the asset management tool, which allows a constant up-dating of information regarding the structural state of the elements and the necessary maintenance activities. Full article
(This article belongs to the Section Intelligent Sensors)
Show Figures

Figure 1

Article
Using Artificial Intelligence to Achieve Auxiliary Training of Table Tennis Based on Inertial Perception Data
Sensors 2021, 21(19), 6685; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196685 - 08 Oct 2021
Viewed by 315
Abstract
Compared with optical sensors, wearable inertial sensors have many advantages such as low cost, small size, more comprehensive application range, no space restrictions and occlusion, better protection of user privacy, and more suitable for sports applications. This article aims to solve irregular actions [...] Read more.
Compared with optical sensors, wearable inertial sensors have many advantages such as low cost, small size, more comprehensive application range, no space restrictions and occlusion, better protection of user privacy, and more suitable for sports applications. This article aims to solve irregular actions that table tennis enthusiasts do not know in actual situations. We use wearable inertial sensors to obtain human table tennis action data of professional table tennis players and non-professional table tennis players, and extract the features from them. Finally, we propose a new method based on multi-dimensional feature fusion convolutional neural network and fine-grained evaluation of human table tennis actions. Realize ping-pong action recognition and evaluation, and then achieve the purpose of auxiliary training. The experimental results prove that our proposed multi-dimensional feature fusion convolutional neural network has an average recognition rate that is 0.17 and 0.16 higher than that of CNN and Inception-CNN on the nine-axis non-professional test set, which proves that we can better distinguish different human table tennis actions and have a more robust generalization performance. Therefore, on this basis, we have better realized the enthusiast of table tennis the purpose of the action for auxiliary training. Full article
(This article belongs to the Special Issue Sensors and AI for Movement Analysis)
Show Figures

Figure 1

Article
A Setup for Measuring the Centering Error of a Dual-Element Pyroelectric Infrared Sensor Module
Sensors 2021, 21(19), 6684; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196684 - 08 Oct 2021
Viewed by 276
Abstract
This paper presents an experimental setup to measure the horizontal centering error of a pre-built pyroelectric infrared (PIR) sensor module, in which a dual-element PIR sensor is aligned at the focal point of a single-zone Fresnel Lens. In the setup, the sensor module [...] Read more.
This paper presents an experimental setup to measure the horizontal centering error of a pre-built pyroelectric infrared (PIR) sensor module, in which a dual-element PIR sensor is aligned at the focal point of a single-zone Fresnel Lens. In the setup, the sensor module was placed facing a modulated infrared radiating source and turned over a range of horizontal angles. The position of the optical axis of the sensor module was determined based on the analysis of the output response of the sensor at turned angles. Thus, the horizontal centering error of the module is defined as the difference between the mechanical axis of the housing and the found optical axis. For the prebuilt sensor module, with the specific available equipment, the measurement of the centering error of the module achieved a resolution of 0.02 degrees. Full article
(This article belongs to the Section Optical Sensors)
Show Figures

Figure 1

Communication
A Frequency-Programmable Miniaturized Radio Frequency Transmitter for Animal Tracking
Sensors 2021, 21(19), 6683; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196683 - 08 Oct 2021
Viewed by 245
Abstract
In animal tracking applications, smaller transmitters can reduce the impact of the transmitter on the tagged animal and thus provide more accurate data about animal behavior. By combining a novel circuit design and a newly developed micro-battery, we developed frequency-programmable and more powerful [...] Read more.
In animal tracking applications, smaller transmitters can reduce the impact of the transmitter on the tagged animal and thus provide more accurate data about animal behavior. By combining a novel circuit design and a newly developed micro-battery, we developed frequency-programmable and more powerful radio frequency transmitters that are about 40% smaller and lighter in weight than the smallest commercial counterpart for animal monitoring at the time of development. The new radio frequency transmitter has a miniaturized form factor for studying small animals. Designs of two coding schemes were developed: one transmits unmodulated signals (weight: 152 mg; dimensions: Ø 2.95 mm × 11.22 mm), and the other transmits modulated signals (weight: 160 mg; dimensions: Ø 2.95 mm × 11.85 mm). To accommodate different transmitter life requirements, each design can be configured to transmit in high or low signal strength. Prototypes of these transmitters were evaluated in the laboratory and exhibited comparable or longer service life and higher signal strength compared to their smallest commercial counterparts. Full article
(This article belongs to the Special Issue Advances and Applications of Micro/Nano-Electronic Sensors)
Show Figures

Figure 1

Article
Improving R Peak Detection in ECG Signal Using Dynamic Mode Selected Energy and Adaptive Window Sizing Algorithm with Decision Tree Algorithm
Sensors 2021, 21(19), 6682; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196682 - 08 Oct 2021
Viewed by 329
Abstract
R peak detection is crucial in electrocardiogram (ECG) signal analysis to detect and diagnose cardiovascular diseases (CVDs). Herein, the dynamic mode selected energy (DMSE) and adaptive window sizing (AWS) algorithm are proposed for detecting R peaks with better efficiency. The DMSE algorithm adaptively [...] Read more.
R peak detection is crucial in electrocardiogram (ECG) signal analysis to detect and diagnose cardiovascular diseases (CVDs). Herein, the dynamic mode selected energy (DMSE) and adaptive window sizing (AWS) algorithm are proposed for detecting R peaks with better efficiency. The DMSE algorithm adaptively separates the QRS components and all non-objective components from the ECG signal. Based on local peaks in QRS components, the AWS algorithm adaptively determines the Region of Interest (ROI). The Feature Extraction process computes the statistical properties of energy, frequency, and noise from each ROI. The Sequential Forward Selection (SFS) procedure is used to find the best subsets of features. Based on these characteristics, an ensemble of decision tree algorithms detects the R peaks. Finally, the R peak position on the initial ECG signal is adjusted using the R location correction (RLC) algorithm. The proposed method has an experimental accuracy of 99.94%, a sensitivity of 99.98%, positive predictability of 99.96%, and a detection error rate of 0.06%. Given the high efficiency in detection and fast processing speed, the proposed approach is ideal for intelligent medical and wearable devices in the diagnosis of CVDs. Full article
(This article belongs to the Special Issue Data Analytics and Applications of Wearable Sensors in Healthcare)
Show Figures

Figure 1

Article
3D Internal Visualization of Concrete Structure Using Multifaceted Data for Ultrasonic Array Pulse-Echo Tomography
Sensors 2021, 21(19), 6681; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196681 - 08 Oct 2021
Viewed by 340
Abstract
This research proposes a 3D internal visualization using ultrasonic pulse-echo tomography technique to evaluate accurately the state of concrete structures for their efficient maintenance within a limited budget. Synthetic aperture focusing technique (SAFT) is used as a post-processing algorithm to manipulate the data [...] Read more.
This research proposes a 3D internal visualization using ultrasonic pulse-echo tomography technique to evaluate accurately the state of concrete structures for their efficient maintenance within a limited budget. Synthetic aperture focusing technique (SAFT) is used as a post-processing algorithm to manipulate the data measured by the ultrasonic pulse-echo technique. Multifaceted measurements improve the weakness of the existing ultrasonic pulse-echo tomography technique that cannot identify the area beyond a reflector as well as the area located far away from measuring surfaces. The application of apodization factor, pulse peak delay calibration and elimination of trivial response not only complements the weaknesses of the SAFT algorithm but also improves the accuracy of the SAFT algorithm. The results show that the proposed method reduces the unnecessary surface noise and improves the expressiveness of the reflector’s boundaries on the resulting images. It is expected that the proposed 3D internal visualization technique will provide a useful non-destructive evaluation tool in combination with another structure evaluation method. Full article
(This article belongs to the Section Sensing and Imaging)
Show Figures

Figure 1

Article
Enhanced Soft 3D Reconstruction Method with an Iterative Matching Cost Update Using Object Surface Consensus
Sensors 2021, 21(19), 6680; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196680 - 08 Oct 2021
Viewed by 300
Abstract
In this paper, we propose a multi-view stereo matching method, EnSoft3D (Enhanced Soft 3D Reconstruction) to obtain dense and high-quality depth images. Multi-view stereo is one of the high-interest research areas and has wide applications. Motivated by the Soft3D reconstruction method, we introduce [...] Read more.
In this paper, we propose a multi-view stereo matching method, EnSoft3D (Enhanced Soft 3D Reconstruction) to obtain dense and high-quality depth images. Multi-view stereo is one of the high-interest research areas and has wide applications. Motivated by the Soft3D reconstruction method, we introduce a new multi-view stereo matching scheme. The original Soft3D method is introduced for novel view synthesis, while occlusion-aware depth is also reconstructed by integrating the matching costs of the Plane Sweep Stereo (PSS) and soft visibility volumes. However, the Soft3D method has an inherent limitation because the erroneous PSS matching costs are not updated. To overcome this limitation, the proposed scheme introduces an update process of the PSS matching costs. From the object surface consensus volume, an inverse consensus kernel is derived, and the PSS matching costs are iteratively updated using the kernel. The proposed EnSoft3D method reconstructs a highly accurate 3D depth image because both the multi-view matching cost and soft visibility are updated simultaneously. The performance of the proposed method is evaluated by using structured and unstructured benchmark datasets. Disparity error is measured to verify 3D reconstruction accuracy, and both PSNR and SSIM are measured to verify the simultaneous enhancement of view synthesis. Full article
(This article belongs to the Special Issue Sensing and Processing for 3D Computer Vision)
Show Figures

Figure 1

Article
Anomaly Detection of Water Level Using Deep Autoencoder
Sensors 2021, 21(19), 6679; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196679 - 08 Oct 2021
Viewed by 295
Abstract
Anomaly detection is one of the crucial tasks in daily infrastructure operations as it can prevent massive damage to devices or resources, which may then lead to catastrophic outcomes. To address this challenge, we propose an automated solution to detect anomaly pattern(s) of [...] Read more.
Anomaly detection is one of the crucial tasks in daily infrastructure operations as it can prevent massive damage to devices or resources, which may then lead to catastrophic outcomes. To address this challenge, we propose an automated solution to detect anomaly pattern(s) of the water levels and report the analysis and time/point(s) of abnormality. This research’s motivation is the level difficulty and time-consuming managing facilities responsible for controlling water levels due to the rare occurrence of abnormal patterns. Consequently, we employed deep autoencoder, one of the types of artificial neural network architectures, to learn different patterns from the given sequences of data points and reconstruct them. Then we use the reconstructed patterns from the deep autoencoder together with a threshold to report which patterns are abnormal from the normal ones. We used a stream of time-series data collected from sensors to train the model and then evaluate it, ready for deployment as the anomaly detection system framework. We run extensive experiments on sensor data from water tanks. Our analysis shows why we conclude vanilla deep autoencoder as the most effective solution in this scenario. Full article
(This article belongs to the Special Issue Machine Learning Applied to Sensor Data Analysis)
Show Figures

Figure 1

Article
Cost-Effective Vibration Analysis through Data-Backed Pipeline Optimisation
Sensors 2021, 21(19), 6678; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196678 - 08 Oct 2021
Viewed by 342
Abstract
Vibration analysis is an active area of research, aimed, among other targets, at an accurate classification of machinery failure modes. The analysis often leads to complex and convoluted signal processing pipeline designs, which are computationally demanding and often cannot be deployed in IoT [...] Read more.
Vibration analysis is an active area of research, aimed, among other targets, at an accurate classification of machinery failure modes. The analysis often leads to complex and convoluted signal processing pipeline designs, which are computationally demanding and often cannot be deployed in IoT devices. In the current work, we address this issue by proposing a data-driven methodology that allows optimising and justifying the complexity of the signal processing pipelines. Additionally, aiming to make IoT vibration analysis systems more cost- and computationally efficient, on the example of MAFAULDA vibration dataset, we assess the changes in the failure classification performance at low sampling rates as well as short observation time windows. We find out that a decrease of the sampling rate from 50 kHz to 1 kHz leads to a statistically significant classification performance drop. A statistically significant decrease is also observed for the 0.1 s time window compared to the 5 s one. However, the effect sizes are small to medium, suggesting that in certain settings lower sampling rates and shorter observation windows might be worth using, consequently making the use of the more cost-efficient sensors feasible. The proposed optimisation approach, as well as the statistically supported findings of the study, allow for an efficient design of IoT vibration analysis systems, both in terms of complexity and costs, bringing us one step closer to the widely accessible IoT/Edge-based vibration analysis. Full article
(This article belongs to the Special Issue Smart Sensors in the Industrial Internet of Things (IIoT))
Show Figures

Figure 1

Article
Investigation of Heterogeneity Sources for Occupational Task Recognition via Transfer Learning
Sensors 2021, 21(19), 6677; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196677 - 08 Oct 2021
Viewed by 305
Abstract
Human activity recognition has been extensively used for the classification of occupational tasks. Existing activity recognition approaches perform well when training and testing data follow an identical distribution. However, in the real world, this condition may be violated due to existing heterogeneities among [...] Read more.
Human activity recognition has been extensively used for the classification of occupational tasks. Existing activity recognition approaches perform well when training and testing data follow an identical distribution. However, in the real world, this condition may be violated due to existing heterogeneities among training and testing data, which results in degradation of classification performance. This study aims to investigate the impact of four heterogeneity sources, cross-sensor, cross-subject, joint cross-sensor and cross-subject, and cross-scenario heterogeneities, on classification performance. To that end, two experiments called separate task scenario and mixed task scenario were conducted to simulate tasks of electrical line workers under various heterogeneity sources. Furthermore, a support vector machine classifier equipped with domain adaptation was used to classify the tasks and benchmarked against a standard support vector machine baseline. Our results demonstrated that the support vector machine equipped with domain adaptation outperformed the baseline for cross-sensor, joint cross-subject and cross-sensor, and cross-subject cases, while the performance of support vector machine equipped with domain adaptation was not better than that of the baseline for cross-scenario case. Therefore, it is of great importance to investigate the impact of heterogeneity sources on classification performance and if needed, leverage domain adaptation methods to improve the performance. Full article
(This article belongs to the Special Issue Data Analytics and Applications of the Wearable Sensors)
Show Figures

Figure 1

Article
Corona Discharge Characteristics under Variable Frequency and Pressure Environments
Sensors 2021, 21(19), 6676; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196676 - 08 Oct 2021
Viewed by 263
Abstract
More electric aircrafts (MEAs) are paving the path to all electric aircrafts (AEAs), which make a much more intensive use of electrical power than conventional aircrafts. Due to the strict weight requirements, both MEA and AEA systems require to increase the distribution voltage [...] Read more.
More electric aircrafts (MEAs) are paving the path to all electric aircrafts (AEAs), which make a much more intensive use of electrical power than conventional aircrafts. Due to the strict weight requirements, both MEA and AEA systems require to increase the distribution voltage in order to limit the required electrical current. Under this paradigm new issues arise, in part due to the voltage rise and in part because of the harsh environments found in aircrafts systems, especially those related to low pressure and high-electric frequency operation. Increased voltage levels, high-operating frequencies, low-pressure environments and reduced distances between wires pose insulation systems at risk, so partial discharges (PDs) and electrical breakdown are more likely to occur. This paper performs an experimental analysis of the effect of low-pressure environments and high-operating frequencies on the visual corona voltage, since corona discharges occurrence is directly related to arc tracking and insulation degradation in wiring systems. To this end, a rod-to-plane electrode configuration is tested in the 20–100 kPa and 50–1000 Hz ranges, these ranges cover most aircraft applications, so that the corona extinction voltage is experimentally determined by using a low-cost high-resolution CMOS imaging sensor which is sensitive to the visible and near ultraviolet (UV) spectra. The imaging sensor locates the discharge points and the intensity of the discharge, offering simplicity and low-cost measurements with high sensitivity. Moreover, to assess the performance of such sensor, the discharges are also acquired by analyzing the leakage current using an inexpensive resistor and a fast oscilloscope. The experimental data presented in this paper can be useful in designing insulation systems for MEA and AEA applications. Full article
(This article belongs to the Special Issue Data, Signal and Image Processing and Applications in Sensors II)
Show Figures

Figure 1

Communication
Elastic Properties Measurement Using Guided Acoustic Waves
Sensors 2021, 21(19), 6675; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196675 - 08 Oct 2021
Viewed by 354
Abstract
Nondestructive evaluation of elastic properties plays a critical role in condition monitoring of thin structures such as sheets, plates or tubes. Recent research has shown that elastic properties of such structures can be determined with remarkable accuracy by utilizing the dispersive nature of [...] Read more.
Nondestructive evaluation of elastic properties plays a critical role in condition monitoring of thin structures such as sheets, plates or tubes. Recent research has shown that elastic properties of such structures can be determined with remarkable accuracy by utilizing the dispersive nature of guided acoustic waves propagating in them. However, existing techniques largely require complicated and expensive equipment or involve accurate measurement of an additional quantity, rendering them impractical for industrial use. In this work, we present a new approach that requires only a pair of piezoelectric transducers used to measure the group velocities ratio of fundamental guided wave modes. A numerical model based on the spectral collocation method is used to fit the measured data by solving a bound-constrained nonlinear least squares optimization problem. We verify our approach on both simulated and experimental data and achieve accuracies similar to those reported by other authors. The high accuracy and simple measurement setup of our approach makes it eminently suitable for use in industrial environments. Full article
(This article belongs to the Section Physical Sensors)
Show Figures

Figure 1

Article
Uncertainty-Aware Knowledge Distillation for Collision Identification of Collaborative Robots
Sensors 2021, 21(19), 6674; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196674 - 08 Oct 2021
Viewed by 273
Abstract
Human-robot interaction has received a lot of attention as collaborative robots became widely utilized in many industrial fields. Among techniques for human-robot interaction, collision identification is an indispensable element in collaborative robots to prevent fatal accidents. This paper proposes a deep learning method [...] Read more.
Human-robot interaction has received a lot of attention as collaborative robots became widely utilized in many industrial fields. Among techniques for human-robot interaction, collision identification is an indispensable element in collaborative robots to prevent fatal accidents. This paper proposes a deep learning method for identifying external collisions in 6-DoF articulated robots. The proposed method expands the idea of CollisionNet, which was previously proposed for collision detection, to identify the locations of external forces. The key contribution of this paper is uncertainty-aware knowledge distillation for improving the accuracy of a deep neural network. Sample-level uncertainties are estimated from a teacher network, and larger penalties are imposed for uncertain samples during the training of a student network. Experiments demonstrate that the proposed method is effective for improving the performance of collision identification. Full article
(This article belongs to the Special Issue Human-Robot Collaborations in Industrial Automation)
Show Figures

Figure 1

Article
Design and Processing Method for Doppler-Tolerant Stepped-Frequency Waveform Using Staggered PRF
Sensors 2021, 21(19), 6673; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196673 - 08 Oct 2021
Viewed by 243
Abstract
Stepped-frequency waveform may be used to synthesize a wideband signal with several narrow-band pulses and achieve a high-resolution range profile without increasing the instantaneous bandwidth. Nevertheless, the conventional stepped-frequency waveform is Doppler sensitive, which greatly limits its application to moving targets. For this [...] Read more.
Stepped-frequency waveform may be used to synthesize a wideband signal with several narrow-band pulses and achieve a high-resolution range profile without increasing the instantaneous bandwidth. Nevertheless, the conventional stepped-frequency waveform is Doppler sensitive, which greatly limits its application to moving targets. For this reason, this paper proposes a waveform design method using a staggered pulse repetition frequency to improve the Doppler tolerance effectively. First, a generalized echo model of the stepped-frequency waveform is constructed in order to analyze the Doppler sensitivity. Then, waveform design is carried out in the stepped-frequency waveform by using a staggered pulse repetition frequency so as to eliminate the high-order phase component that is caused by the target’s velocity. Further, the waveform design method is extended to the sparse stepped-frequency waveform, and we also propose corresponding methods for high-resolution range profile synthesis and motion compensation. Finally, experiments with electromagnetic data verify the high Doppler tolerance of the proposed waveform. Full article
(This article belongs to the Section Radar Sensors)
Show Figures

Figure 1

Article
Multi-Domain Convolutional Neural Networks for Lower-Limb Motor Imagery Using Dry vs. Wet Electrodes
Sensors 2021, 21(19), 6672; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196672 - 07 Oct 2021
Viewed by 435
Abstract
Motor imagery (MI) brain–computer interfaces (BCIs) have been used for a wide variety of applications due to their intuitive matching between the user’s intentions and the performance of tasks. Applying dry electroencephalography (EEG) electrodes to MI BCI applications can resolve many constraints and [...] Read more.
Motor imagery (MI) brain–computer interfaces (BCIs) have been used for a wide variety of applications due to their intuitive matching between the user’s intentions and the performance of tasks. Applying dry electroencephalography (EEG) electrodes to MI BCI applications can resolve many constraints and achieve practicality. In this study, we propose a multi-domain convolutional neural networks (MD-CNN) model that learns subject-specific and electrode-dependent EEG features using a multi-domain structure to improve the classification accuracy of dry electrode MI BCIs. The proposed MD-CNN model is composed of learning layers for three domain representations (time, spatial, and phase). We first evaluated the proposed MD-CNN model using a public dataset to confirm 78.96% classification accuracy for multi-class classification (chance level accuracy: 30%). After that, 10 healthy subjects participated and performed three classes of MI tasks related to lower-limb movement (gait, sitting down, and resting) over two sessions (dry and wet electrodes). Consequently, the proposed MD-CNN model achieved the highest classification accuracy (dry: 58.44%; wet: 58.66%; chance level accuracy: 43.33%) with a three-class classifier and the lowest difference in accuracy between the two electrode types (0.22%, d = 0.0292) compared with the conventional classifiers (FBCSP, EEGNet, ShallowConvNet, and DeepConvNet) that used only a single domain. We expect that the proposed MD-CNN model could be applied for developing robust MI BCI systems with dry electrodes. Full article
(This article belongs to the Special Issue EEG-Based Brain–Computer Interface for a Real-Life Appliance)
Show Figures

Figure 1

Article
Photothermocapillary Method for the Nondestructive Testing of Solid Materials and Thin Coatings
Sensors 2021, 21(19), 6671; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196671 - 07 Oct 2021
Viewed by 354
Abstract
The photothermocapillary (PTC) effect is a deformation of the free surface of a thin liquid layer on a solid material that is caused by the dependence of the coefficient of surface tension on temperature. The PTC effect is highly sensitive to variations in [...] Read more.
The photothermocapillary (PTC) effect is a deformation of the free surface of a thin liquid layer on a solid material that is caused by the dependence of the coefficient of surface tension on temperature. The PTC effect is highly sensitive to variations in the thermal conductivity of solids, and this is the basis for PTC techniques in the non-destructive testing of solid non-porous materials. These techniques analyze thermal conductivity and detect subsurface defects, evaluate the thickness of thin varnish-and-paint coatings (VPC), and detect air-filled voids between coatings and metal substrates. In this study, the PTC effect was excited by a “pumped” Helium-Neon laser, which provided the monochromatic light source that is required to produce optical interference patterns. The light of a small-diameter laser beam was reflected from a liquid surface, which was contoured by liquid capillary action and variations in the surface tension. A typical contour produces an interference pattern of concentric rings with a bright and wide outer ring. The minimal or maximal diameter of this pattern was designated as the PTC response. The PTC technique was evaluated to monitor the thickness of VPCs on thermally conductive solid materials. The same PTC technique has been used to measure the thickness of air-filled delaminations between a metal substrate and a coating. Full article
(This article belongs to the Topic Fiber-Reinforced Cementitious Composites)
Show Figures

Figure 1

Article
Speed Management Strategy: Designing an IoT-Based Electric Vehicle Speed Control Monitoring System
Sensors 2021, 21(19), 6670; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196670 - 07 Oct 2021
Viewed by 458
Abstract
Road accidents represent the greatest public health burden in the world. Road traffic accidents have been on the rise in Rwanda for several years. Speed has been identified as a core factor in these road accidents. Therefore, understanding road accidents caused by excessive [...] Read more.
Road accidents represent the greatest public health burden in the world. Road traffic accidents have been on the rise in Rwanda for several years. Speed has been identified as a core factor in these road accidents. Therefore, understanding road accidents caused by excessive speeding is critical for road safety planning. In this paper, input and out pulse width modulation (PWM) was used to command the metal–oxide–semiconductor field-effect transistor (MOSFET) controller which supplied voltage to the motor. A structural speed control and Internet of Things (IoT)-based online monitoring system was developed to monitor vehicle data in a continuous manner. Two modeling techniques, multiple linear regression (MLR) and random forest (RF) models, were evaluated to find the best model to estimate the required voltage to be supplied to the motors in a particular zone. The built models were evaluated based upon the coefficient of determination R2. The RF performs better than the MLR as it reveals a higher R2 value and it is found to be 98.8%. Based on the results, the proposed method was proven to significantly reduce the supplied voltage to the motor and consequently increase safety. Full article
(This article belongs to the Special Issue Sensors Technologies for Intelligent Transportation Systems)
Show Figures

Figure 1

Article
DATSURYOKU Sensor—A Capacitive-Sensor-Based Belt for Predicting Muscle Tension: Preliminary Results
Sensors 2021, 21(19), 6669; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196669 - 07 Oct 2021
Viewed by 518
Abstract
Excessive muscle tension is implicitly caused by inactivity or tension in daily activities, and it results in increased joint stiffness and vibration, and thus, poor performance, failure, and injury in sports. Therefore, the routine measurement of muscle tension is important. However, a co-contraction [...] Read more.
Excessive muscle tension is implicitly caused by inactivity or tension in daily activities, and it results in increased joint stiffness and vibration, and thus, poor performance, failure, and injury in sports. Therefore, the routine measurement of muscle tension is important. However, a co-contraction observed in excessive muscle tension cannot be easily detected because it does not appear in motion owing to the counteracting muscle tension, and it cannot be measured by conventional motion capture systems. Therefore, we focused on the physiological characteristics of muscle, that is, the increase in muscle belly cross-sectional area during activity and softening during relaxation. Furthermore, we measured muscle tension, especially co-contraction and relaxation, using a DATSURYOKU sensor, which measures the circumference of the applied part. The experiments showed high interclass correlation between muscle activities and circumference across maximal voluntary co-contractions of the thigh muscles and squats. Moreover, the circumference sensor can measure passive muscle deformation that does not appear in muscle activities. Therefore, the DATSURYOKU sensor showed the potential to routinely measure muscle tension and relaxation, thus avoiding the risk of failure and injury owing to excessive muscle tension and can contribute to the realization of preemptive medicine by measuring daily changes. Full article
(This article belongs to the Special Issue Wearable Sensors for Risk Assessment and Injury Prevention)
Show Figures

Figure 1

Article
On-Orbit Geometric Calibration from the Relative Motion of Stars for Geostationary Cameras
Sensors 2021, 21(19), 6668; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196668 - 07 Oct 2021
Viewed by 274
Abstract
Affected by the vibrations and thermal shocks during launch and the orbit penetration process, the geometric positioning model of the remote sensing cameras measured on the ground will generate a displacement, affecting the geometric accuracy of imagery and requiring recalibration. Conventional methods adopt [...] Read more.
Affected by the vibrations and thermal shocks during launch and the orbit penetration process, the geometric positioning model of the remote sensing cameras measured on the ground will generate a displacement, affecting the geometric accuracy of imagery and requiring recalibration. Conventional methods adopt the ground control points (GCPs) or stars as references for on-orbit geometric calibration. However, inescapable cloud coverage and discontented extraction algorithms make it extremely difficult to collect sufficient high-precision GCPs for modifying the misalignment of the camera, especially for geostationary satellites. Additionally, the number of the observed stars is very likely to be inadequate for calibrating the relative installations of the camera. In terms of the problems above, we propose a novel on-orbit geometric calibration method using the relative motion of stars for geostationary cameras. First, a geometric calibration model is constructed based on the optical system structure. Then, we analyze the relative motion transformation of the observed stars. The stellar trajectory and the auxiliary ephemeris are used to obtain the corresponding object vector for correcting the associated calibration parameters iteratively. Experimental results evaluated on the data of a geostationary experiment satellite demonstrate that the positioning errors corrected by this proposed method can be within ±2.35 pixels. This approach is able to effectively calibrate the camera and improve the positioning accuracy, which avoids the influence of cloud cover and overcomes the great dependence on the number of the observed stars. Full article
(This article belongs to the Section Navigation and Positioning)
Show Figures

Figure 1

Article
Evaluating the Dynamics of Bluetooth Low Energy Based COVID-19 Risk Estimation for Educational Institutes
Sensors 2021, 21(19), 6667; https://0-doi-org.brum.beds.ac.uk/10.3390/s21196667 - 07 Oct 2021
Viewed by 312
Abstract
COVID-19 tracing applications have been launched in several countries to track and control the spread of viruses. Such applications utilize Bluetooth Low Energy (BLE) transmissions, which are short range and can be used to determine infected and susceptible persons near an infected person. [...] Read more.
COVID-19 tracing applications have been launched in several countries to track and control the spread of viruses. Such applications utilize Bluetooth Low Energy (BLE) transmissions, which are short range and can be used to determine infected and susceptible persons near an infected person. The COVID-19 risk estimation depends on an epidemic model for the virus behavior and Machine Learning (ML) model to classify the risk based on time series distance of the nodes that may be infected. The BLE technology enabled smartphones continuously transmit beacons and the distance is inferred from the received signal strength indicators (RSSI). The educational activities have shifted to online teaching modes due to the contagious nature of COVID-19. The government policy makers decide on education mode (online, hybrid, or physical) with little technological insight on actual risk estimates. In this study, we analyze BLE technology to debate the COVID-19 risks in university block and indoor class environments. We utilize a sigmoid based epidemic model with varying thresholds of distance to label contact data with high risk or low risk based on features such as contact duration. Further, we train multiple ML classifiers to classify a person into high risk or low risk based on labeled data of RSSI and distance. We analyze the accuracy of the ML classifiers in terms of F-score, receiver operating characteristic (ROC) curve, and confusion matrix. Lastly, we debate future research directions and limitations of this study. We complement the study with open source code so that it can be validated and further investigated. Full article
Show Figures

Figure 1

Previous Issue
Back to TopTop