A Novel Method for Cross-Subject Human Activity Recognition with Wearable Sensors

Human Activity Recognition (HAR) is an important way for lower limb exoskeleton robots to implement human-computer collaboration with users. Most of the existing methods in this field focus on a simple scenario recognizing activities for specific users, which does not consider the individual differences among users and cannot adapt to new users. In order to improve the generalization ability of HAR model, this paper proposes a novel method that combines the theories in transfer learning and active learning to mitigate the cross-subject issue, so that it can enable lower limb exoskeleton robots being used in more complex scenarios. First, a neural network based on convolutional neural networks (CNN) is designed, which can extract temporal and spatial features from sensor signals collected from different parts of human body. It can recognize human activities with high accuracy after trained by labeled data. Second, in order to improve the cross-subject adaptation ability of the pre-trained model, we design a cross-subject HAR algorithm based on sparse interrogation and label propagation. Through leave-one-subject-out validation on two widely-used public datasets with existing methods, our method achieves average accuracies of 91.77% on DSAD and 80.97% on PAMAP2, respectively. The experimental results demonstrate the potential of implementing cross-subject HAR for lower limb exoskeleton robots.

Deep Pyramidal Residual Network for Indoor-Outdoor Activity Recognition Based on Wearable Sensor

Recognition of human activity is one of the most exciting aspects of time-series classification,with substantial practical and theoretical impli-cations.Recent evidence indicates that activity recognition from wearable sensors is an effective technique for tracking elderly adults and children in indoor and outdoor environments.Consequently,researchers have demon-strated considerable passion for developing cutting-edge deep learning sys-tems capable of exploiting unprocessed sensor data from wearable devices and generating practical decision assistance in many contexts.This study provides a deep learning-based approach for recognizing indoor and outdoor movement utilizing an enhanced deep pyramidal residual model called Sen-PyramidNet and motion information from wearable sensors(accelerometer and gyroscope).The suggested technique develops a residual unit based on a deep pyramidal residual network and introduces the concept of a pyramidal residual unit to increase detection capability.The proposed deep learning-based model was assessed using the publicly available 19Nonsens dataset,which gathered motion signals from various indoor and outdoor activities,including practicing various body parts.The experimental findings demon-strate that the proposed approach can efficiently reuse characteristics and has achieved an identification accuracy of 96.37%for indoor and 97.25%for outdoor activity.Moreover,comparison experiments demonstrate that the SenPyramidNet surpasses other cutting-edge deep learning models in terms of accuracy and F1-score.Furthermore,this study explores the influence of several wearable sensors on indoor and outdoor action recognition ability.

Silk Fibroin-Based Hydrogel for Multifunctional Wearable Sensors

The flexible wearable sensors with excellent stretchability,high sensitivity and good biocompatibility are significantly required for continuously physical condition tracking in health management and rehabilitation monitoring.Herein,we present a high-performance wearable sensor.The sensor is prepared with nanocomposite hydrogel by using silk fibroin(SF),polyacrylamide(PAM),polydopamine(PDA)and graphene oxide(GO).It can be used to monitor body motions(including large-scale and small-scale motions)as well as human electrophysiological(ECG)signals with high sensitivity,wide sensing range,and fast response time.Therefore,the proposed sensor is promising in the fields of rehabilitation,motion monitoring and disease diagnosis.

Flexible wearable sensors:An emerging platform for monitoring of bacterial infection in skin wounds

Persistent inflammatory responses often occur when bacteria and other microorganisms frequently invade and colonize open wounds and eventually result in the formation of chronic wounds.Therefore,achieving real-time detection of invasive bacteria accurately and promptly is essential for efficient wound management and accelerat-ing the healing process.Recently,flexible wearable sensors have garnered significant attention,especially those designed for monitoring real-time biophysical or biochemical signals in wound sites in a minimally invasive manner.They provide more precise and continuous monitoring data,making them as emerging tools for clinical diagnostics.In this review,we first discuss the species and community distribution of different types of bacteria in chronic wounds.Next,we introduce currently developed techniques for detecting bacteria at wound sites.Fol-lowing that,we discuss the recent progress and unresolved issues of various flexible wearable sensors in detecting bacteria at wound sites.We believe that this review can provide meaningful guidance for the development of flexible wearable sensors for bacteria detection.

Hybrid pedestrian positioning system using wearable inertial sensors and ultrasonic ranging

Pedestrian positioning system(PPS)using wearable inertial sensors has wide applications towards various emerging fields such as smart healthcare,emergency rescue,soldier positioning,etc.The performance of traditional PPS is limited by the cumulative error of inertial sensors,complex motion modes of pedestrians,and the low robustness of the multi-sensor collaboration structure.This paper presents a hybrid pedestrian positioning system using the combination of wearable inertial sensors and ultrasonic ranging(H-PPS).A robust two nodes integration structure is developed to adaptively combine the motion data acquired from the single waist-mounted and foot-mounted node,and enhanced by a novel ellipsoid constraint model.In addition,a deep-learning-based walking speed estimator is proposed by considering all the motion features provided by different nodes,which effectively reduces the cumulative error originating from inertial sensors.Finally,a comprehensive data and model dual-driven model is presented to effectively combine the motion data provided by different sensor nodes and walking speed estimator,and multi-level constraints are extracted to further improve the performance of the overall system.Experimental results indicate that the proposed H-PPS significantly improves the performance of the single PPS and outperforms existing algorithms in accuracy index under complex indoor scenarios.

Functionalized Hydrogel-Based Wearable Gas and Humidity Sensors

Breathing is an inherent human activity;however,the composition of the air we inhale and gas exhale remains unknown to us.To address this,wearable vapor sensors can help people monitor air composition in real time to avoid underlying risks,and for the early detection and treatment of diseases for home healthcare.Hydrogels with three-dimensional polymer networks and large amounts of water molecules are naturally flexible and stretchable.Functionalized hydrogels are intrinsically conductive,self-healing,self-adhesive,biocompatible,and room-temperature sensitive.Compared with traditional rigid vapor sensors,hydrogel-based gas and humidity sensors can directly fit human skin or clothing,and are more suitable for real-time monitoring of personal health and safety.In this review,current studies on hydrogel-based vapor sensors are investigated.The required properties and optimization methods of wearable hydrogel-based sensors are introduced.Subsequently,existing reports on the response mechanisms of hydrogel-based gas and humidity sensors are summarized.Related works on hydrogel-based vapor sensors for their application in personal health and safety monitoring are presented.Moreover,the potential of hydrogels in the field of vapor sensing is elucidated.Finally,the current research status,challenges,and future trends of hydrogel gas/humidity sensing are discussed.

Wearable sensors for activity monitoring and motion control:A review

Wearable sensors for activity monitoring currently are being designed and developed,driven by an increasing demand in health care for noninvasive patient monitoring and rehabilitation training.This article reviews state-of-the-art wearable sensors for activity monitoring and motion control.Different technologies,including electromechanical,bioelectrical,and biomechanical sensors,are reviewed,along with their broad applications.Moreover,an overview of existing commercial wearable products and the computation methods for motion analysis are provided.Future research issues are identified and discussed.

Wearable and stretchable conductive polymer composites for strain sensors:How to design a superior one?

Wearable and stretchable strain sensors have potential values in the fields of human motion and health monitoring,flexible electronics,and soft robotic skin.The wearable and stretchable strain sensors can be directly attached to human skin,providing visualized detection for human motions and personal healthcare.Conductive polymer composites(CPC)composed of conductive fillers and flexible polymers have the advantages of high stretchability,good flexibility,superior durability,which can be used to prepare flexible strain sensors with large working strain and outstanding sensitivity.This review has put forward a comprehensive summary on the fabrication methods,advanced mechanisms and strain sensing abilities of CPC strain sensors reported in recent years,especially the sensors with superior performance.Finally,the structural design,bionic function,integration technology and further application of CPC strain sensors are prospected.

Automatic Recognition of Construction Worker Activities Using Deep Learning Approaches and Wearable Inertial Sensors

The automated evaluation and analysis of employee behavior in an Industry 4.0-compliant manufacturingfirm are vital for the rapid and accurate diagnosis of work performance,particularly during the training of a new worker.Various techniques for identifying and detecting worker performance in industrial applications are based on computer vision techniques.Despite widespread com-puter vision-based approaches,it is challenging to develop technologies that assist the automated monitoring of worker actions at external working sites where cam-era deployment is problematic.Through the use of wearable inertial sensors,we propose a deep learning method for automatically recognizing the activities of construction workers.The suggested method incorporates a convolutional neural network,residual connection blocks,and multi-branch aggregate transformation modules for high-performance recognition of complicated activities such as con-struction worker tasks.The proposed approach has been evaluated using standard performance measures,such as precision,F1-score,and AUC,using a publicly available benchmark dataset known as VTT-ConIoT,which contains genuine con-struction work activities.In addition,standard deep learning models(CNNs,RNNs,and hybrid models)were developed in different empirical circumstances to compare them to the proposed model.With an average accuracy of 99.71%and an average F1-score of 99.71%,the experimentalfindings revealed that the suggested model could accurately recognize the actions of construction workers.Furthermore,we examined the impact of window size and sensor position on the identification efficiency of the proposed method.

On-body propagation characterization based on FDTD method for 2.4/5.2/5.7 GHz wearable body sensor networks

The on-body path loss and time delay of radio propagation in 2. 4/5.2/5.7 GHz wearable body sensor networks （W-BSN） are studied using Remcom XFDTD, a simulation tool based on the finite-difference time- domain method. The simulation is performed in the environment of free space with a simplified three- dimensional human body model. Results show that the path loss at a higher radio frequency is significantly smaller. Given that the transmitter and the receiver are located on the body trunk, the path loss relevant to the proposed minimum equivalent surface distance follows a log-fitting parametric model, and the path loss exponents are 4. 7, 4. 1 and 4. 0 at frequencies of 2. 4, 5.2, 5.7 GHz, respectively. On the other hand, the first- arrival delays are less than 2 ns at all receivers, and the maximum time delay spread is about 10 ns. As suggested by the maximum time delay spread, transmission rates of W-BSN must be less than 10^8 symbol/s to avoid intersymbol interference from multiple-path delay.

One-step growth of large-area silicon nanowire fabrics for high-performance multifunctional wearable sensors

Silicon nanowire(SiNW)fabrics are of great interest for fabricating high-performance multifunctional wearable sensors.However,it remains a big challenge to fabricate high-quality SiNW fabrics in a simple and efficient manner.Here we report,for the first time,one-step growth of large-area SiNW fabrics for multifunctional wearable sensors,by using a massive metal-assisted chemical vapor deposition(CVD)method.With bulk Sn as a catalyst source,numerous millimeter-long SiNWs grow and naturally interweave with each other,forming SiNw fabrics over 80 cm2 in one experiment.In addition to intrinsic electronic properties of Si materials,the SiNw fabrics also feature high flxibility,good tailorability and light weight,rendering them ideal for fabricating multifunctional wearable sensors.The prototype sensors based on the SiNW fabrics could efectively detect various stimuli including temperature,light,strain and pressure,with outstanding performance among reported multifunctional sensors.We further demonstrate the integration of the prototype sensors onto the body of a robot,enabling its perception to various environmental stimuli.The ability to prepare high-quality SiNW fabrics in a simple and eficient manner will stimulate the development of wearable devices for applications in portable electronics,Internet of Things,health care and robotics.

Laser direct writing of Ga_(2)O_(3)/liquid metal-based flexible humidity sensors

Flexible and wearable humidity sensors play a vital role in daily point-of-care diagnosis and noncontact human-machine interactions.However,achieving a facile and high-speed fabrication approach to realizing flexible humidity sensors remains a challenge.In this work,a wearable capacitive-type Ga_(2)O_(3)/liquid metal-based humidity sensor is demonstrated by a one-step laser direct writing technique.Owing to the photothermal effect of laser,the Ga_(2)O_(3)-wrapped liquid metal particles can be selectively sintered and converted from insulative to conductive traces with a resistivity of 0.19Ω·cm,while the untreated regions serve as active sensing layers in response to moisture changes.Under 95%relative humidity,the humidity sensor displays a highly stable performance along with rapid response and recover time.Utilizing these superior properties,the Ga_(2)O_(3)/liquid metal-based humidity sensor is able to monitor human respiration rate,as well as skin moisture of the palm under different physiological states for healthcare monitoring.

Development of smart wearable sensors for life healthcare

Wearable biosensors are gaining tremendous interest in the clinical and biological medical fields for their potential in providing patients with real-time diagnostic tools and time-sensitive information,non-invasive measurements of biochemical markers distributed in body fluids throughout the body.These sensors replace part of the equipment that can only be installed in hospitals and become a new choice for assessing human health.Herein,critical perspectives are put forward regarding wearable sensors in the future digital health monitoring industry.Moreover,the materials and processing technologies involved have also been discussed in recent years,along with their monitoring schemes and system-level integration technologies.Finally,the probability of wearable sensors being used in early disease detection is considerable.Thus,despite potential challenges,it is still quite promising for wearable sensors to come into production.

Engineering Smart Composite Hydrogels for Wearable Health Monitoring

Growing health awareness triggers the public's concern about health problems. People want a timely and comprehensive picture of their condition without frequent trips to the hospital for costly and cumbersome general check-ups. The wearable technique provides a continuous measurement method for health monitoring by tracking a person's physiological data and analyzing it locally or remotely.During the health monitoring process,different kinds of sensors convert physiological signals into electrical or optical signals that can be recorded and transmitted, consequently playing a crucial role in wearable techniques. Wearable application scenarios usually require sensors to possess excellent flexibility and stretchability. Thus, designing flexible and stretchable sensors with reliable performance is the key to wearable technology. Smart composite hydrogels, which have tunable electrical properties, mechanical properties, biocompatibility, and multi-stimulus sensitivity, are one of the best sensitive materials for wearable health monitoring. This review summarizes the common synthetic and performance optimization strategies of smart composite hydrogels and focuses on the current application of smart composite hydrogels in the field of wearable health monitoring.

Pre-Impact and Impact Fall Detection Based on a Multimodal Sensor Using a Deep Residual Network

Falls are the contributing factor to both fatal and nonfatal injuries in the elderly.Therefore,pre-impact fall detection,which identifies a fall before the body collides with the floor,would be essential.Recently,researchers have turned their attention from post-impact fall detection to pre-impact fall detection.Pre-impact fall detection solutions typically use either a threshold-based or machine learning-based approach,although the threshold value would be difficult to accu-rately determine in threshold-based methods.Moreover,while additional features could sometimes assist in categorizing falls and non-falls more precisely,the esti-mated determination of the significant features would be too time-intensive,thus using a significant portion of the algorithm’s operating time.In this work,we developed a deep residual network with aggregation transformation called FDSNeXt for a pre-impact fall detection approach employing wearable inertial sensors.The proposed network was introduced to address the limitations of fea-ture extraction,threshold definition,and algorithm complexity.After training on a large-scale motion dataset,the KFall dataset,and straightforward evaluation with standard metrics,the proposed approach identified pre-impact and impact falls with high accuracy of 91.87 and 92.52%,respectively.In addition,we have inves-tigated fall detection’s performances of three state-of-the-art deep learning models such as a convolutional neural network(CNN),a long short-term memory neural network(LSTM),and a hybrid model(CNN-LSTM).The experimental results showed that the proposed FDSNeXt model outperformed these deep learning models(CNN,LSTM,and CNN-LSTM)with significant improvements.

Driving Activity Classification Using Deep Residual Networks Based on Smart Glasses Sensors

Accidents are still an issue in an intelligent transportation system,despite developments in self-driving technology(ITS).Drivers who engage in risky behavior account for more than half of all road accidents.As a result,reckless driving behaviour can cause congestion and delays.Computer vision and multimodal sensors have been used to study driving behaviour categorization to lessen this problem.Previous research has also collected and analyzed a wide range of data,including electroencephalography(EEG),electrooculography(EOG),and photographs of the driver’s face.On the other hand,driving a car is a complicated action that requires a wide range of body move-ments.In this work,we proposed a ResNet-SE model,an efficient deep learning classifier for driving activity clas-sification based on signal data obtained in real-world traffic conditions using smart glasses.End-to-end learning can be achieved by combining residual networks and channel attention approaches into a single learning model.Sensor data from 3-point EOG electrodes,tri-axial accelerometer,and tri-axial gyroscope from the Smart Glasses dataset was utilized in this study.We performed various experiments and compared the proposed model to base-line deep learning algorithms(CNNs and LSTMs)to demonstrate its performance.According to the research results,the proposed model outperforms the previous deep learning models in this domain with an accuracy of 99.17%and an F1-score of 98.96%.

Functionalized Fiber-Based Strain Sensors:Pathway to Next-Generation Wearable Electronics

Wearable strain sensors are arousing increasing research interests in recent years on account of their potentials in motion detection,personal and public healthcare,future entertainment,man-machine interaction,artificial intelligence,and so forth.Much research has focused on fiber-based sensors due to the appealing performance of fibers,including processing flexibility,wearing comfortability,outstanding lifetime and serviceability,low-cost and large-scale capacity.Herein,we review the latest advances in functionalization and device fabrication of fiber materials toward applications in fiber-based wearable strain sensors.We describe the approaches for preparing conductive fibers such as spinning,surface modification,and structural transformation.We also introduce the fabrication and sensing mechanisms of state-of-the-art sensors and analyze their merits and demerits.The applications toward motion detection,healthcare,man-machine interaction,future entertainment,and multifunctional sensing are summarized with typical examples.We finally critically analyze tough challenges and future remarks of fiber-based strain sensors,aiming to implement them in real applications.

Advances in self-powered sports monitoring sensors based on triboelectric nanogenerators

The new era of the internet of things brings great opportunities to the field of intelligent sports.The collection and analysis of sports data are becoming more intelligent driven by the widely-distributed sensing network system.Triboelectric nanogenerators(TENGs)can collect and convert energy as selfpowered sensors,overcoming the limitations of external power supply,frequent power replacement and high-cost maintenance.Herein,we introduce the working modes and principles of TENGs,and then summarize the recent advances in self-powered sports monitoring sensors driven by TENGs in sports equipment facilities,wearable equipment and competitive sports specialities.We discuss the existing issues,i.e.,device stability,material sustainability,device design rationality,textile TENG cleanability,sports sensors safety,kinds and manufacturing of sports sensors,and data collection comprehensiveness,and finally,propose the countermeasures.This work has practical significance to the current TENG applications in sports monitoring,and TENG-based sensing technology will have a broad prospect in the field of intelligent sports in the future.

Ultrasensitive stretchable patches for joint motion monitoring

Wearable devices have great application potential in the next generation of smart portable electronics,especially in the fields of medical monitoring,soft robotics,artificial intelligence,and human-machine interfaces.Piezoelectric flexible strain sensors are key components of wearable devices.However,existing piezoelectric flexible strain sensors have certain limitations in weak signal monitoring due to their large modulus and low sensitivity.To solve this problem,the concept of Kirigami(paper-cutting)was introduced in this study to design the sensor structure.By comparing the Kirigami structures of different basic structures,the serpentine structure was determined as the basic configuration of the sensor.The serpentine structure not only provides excellent tensile properties,but also significantly improves the sensitivity of the sensor,which performs well in monitoring weak signals.On this basis,the adhesion properties of the flexible sensor were analyzed and tested,and the optimal ratio of the substrate was selected for preparation.In addition,a low-cost and rapid prototyping process for stretchable patches was established in this study.Using this technology,we prepared the sensor device and tested its performance.Finally,we successfully developed a flexible sensor with a sensitivity of 0.128 mV/μɛand verified its feasibility for wrist joint motion monitoring applications.This result opens up new avenues for the recovery care of tenosynovitis patients after surgery.

Multifunctional flexible optical waveguide sensor: on the bioinspiration for ultrasensitive sensors development

This paper presents the development of a bioinspired multifunctional flexible optical sensor(BioMFOS)as an ultrasensitive tool for force(intensity and location)and orientation sensing.The sensor structure is bioinspired in orb webs,which are multifunctional devices for prey capturing and vibration transmission.The multifunctional feature of the structure is achieved by using transparent resins that present both mechanical and optical properties for structural integrity and strain/deflection transmission as well as the optical signal transmission properties with core/cladding configuration of a waveguide.In this case,photocurable and polydimethylsiloxane(PDMS)resins are used for the core and cladding,respectively.The optical transmission,tensile tests,and dynamic mechanical analysis are performed in the resins and show the possibility of light transmission at the visible wavelength range in conjunction with high flexibility and a dynamic range up to 150 Hz,suitable for wearable applications.The BioMFOS has small dimensions(around 2 cm)and lightweight(0.8 g),making it suitable for wearable application and clothing integration.Characterization tests are performed in the structure by means of applying forces at different locations of the structure.The results show an ultra-high sensitivity and resolution,where forces in theμN range can be detected and the location of the applied force can also be detected with a sub-millimeter spatial resolution.Then,the BioMFOS is tested on the orientation detection in 3D plane,where a correlation coefficient higher than 0.9 is obtained when compared with a gold-standard inertial measurement unit(IMU).Furthermore,the device also shows its capabilities on the movement analysis and classification in two protocols:finger position detection(with the BioMFOS positioned on the top of the hand)and trunk orientation assessment(with the sensor integrated on the clothing).In both cases,the sensor is able of classifying the movement,especially when analyzed in conjunction with preprocessing and clustering techniques.As another wearable application,the respiratory rate is successfully estimated with the BioMFOS integrated into the clothing.Thus,the proposed multifunctional device opens new avenues for novel bioinspired photonic devices and can be used in many applications of biomedical,biomechanics,and micro/nanotechnology.