Search

Abstract Physiological signal monitoring and driver behavior analysis have gained increasing attention in both fundamental research and applied research. This study involved the analysis of driving behavior using multimodal physiological data collected from 35 participants. The data included 59-channel EEG, single-channel ECG, 4-channel EMG, single-channel GSR, and eye movement data obtained via a six-degree-of-freedom driving simulator. We categorized driving behavior into five groups: smooth driving, acceleration, deceleration, lane changing, and turning. Through extensive experiments, we confirmed that both physiological and vehicle data met the requirements. Subsequently, we developed classification models, including linear discriminant analysis (LDA), MMPNet, and EEGNet, to demonstrate the correlation between physiological data and driving behaviors. Notably, we propose a multimodal physiological dataset for analyzing driving behavior(MPDB). The MPDB dataset’s scale, accuracy, and multimodality provide unprecedented opportunities for researchers in the autonomous driving field and beyond. With this dataset, we will contribute to the field of traffic psychology and behavior.

Abstract Moisture-electric generators (MEGs) has emerged as promising green technology to achieve carbon neutrality in next-generation energy suppliers, especially combined with ecofriendly materials. Hitherto, challenges remain for MEGs as direct power source in practical applications due to low and intermittent electric output. Here we design a green MEG with high direct-current electricity by introducing polyvinyl alcohol-sodium alginate-based supramolecular hydrogel as active material. A single unit can generate an improved power density of ca. 0.11 mW cm−2, a milliamp-scale short-circuit current density of ca. 1.31 mA cm−2 and an open-circuit voltage of ca. 1.30 V. Such excellent electricity is mainly attributed to enhanced moisture absorption and remained water gradient to initiate ample ions transport within hydrogel by theoretical calculation and experiments. Notably, an enlarged current of ca. 65 mA is achieved by a parallel-integrated MEG bank. The scalable MEGs can directly power many commercial electronics in real-life scenarios, such as charging smart watch, illuminating a household bulb, driving a digital clock for one month. This work provides new insight into constructing green, high-performance and scalable energy source for Internet-of-Things and wearable applications.

The electroencephalogram-based (EEG) brain-computer interface (BCI) has garnered significant attention in recent research. However, the practicality of EEG remains constrained by the lack of efficient EEG decoding technology. The challenge lies in effectively translating intricate EEG into meaningful, generalizable information. EEG signal decoding primarily relies on either time domain or frequency domain information. There lacks a method capable of simultaneously and effectively extracting both time and frequency domain features, as well as efficiently fuse these features. Addressing these limitations, a two-branch Manifold Domain enhanced transformer algorithm is designed to holistically capture EEG’s spatio-temporal information. Our method projects the time-domain information of EEG signals into the Riemannian spaces to fully decode the time dependence of EEG signals. Using wavelet transform, the time domain information is converted into frequency domain information, and the spatial information contained in the frequency domain information of EEG signal is mined through the spectrogram. The effectiveness of the proposed TBEEG algorithm is validated on BCIC-IV-2a dataset and MAMEM-SSVEP-II datasets.

Current mainstream computer vision algorithms focus on designing suitable network architectures and loss functions to fit training data. However, the accuracy of small object detection remains lower than for other scales, and the design of convolution operators limits the model’s performance. For UAV small object detection, standard convolutions, due to their fixed kernel size, cannot adaptively capture small object spatial information. Many convolutional variants have scattered sampling points, leading to blurred boundaries and reduced accuracy. In response, we propose HawkEye Conv (HEConv), which utilizes stable sampling and dynamic offsets with random selection. By varying the convolution kernel design, HEConv reduces the accuracy gap between small and larger objects while offering multiple versions and plug-and-play capabilities. We also develop HawkEye Spatial Pyramid Pooling and Gradual Dynamic Feature Pyramid Network modules to validate HEConv. Experiments on the RFRB agricultural and VisDrone2019 urban datasets demonstrate that, compared to YOLOv10, our model improves AP50 by 11.9% and 6.2%, APS by 11.5% and 5%, and F1-score by 5% and 7%. Importantly, it enhances small object detection without sacrificing large object accuracy, thereby reducing the multi-scale performance gap.

Addressing the challenges in online multi-object tracking algorithms under complex scenarios, where the independence among feature extraction, object detection, and data association modules leads to both error accumulation and the difficulty of maintaining visual consistency for occluded objects, we have proposed an end-to-end multi-object tracking method based on hypergraph matching (JDTHM). Initially, a feature extraction and object detection module is introduced to achieve preliminary localization and description of the objects. Subsequently, a deep feature aggregation module is designed to extract temporal information from historical tracklets, amalgamating features from object detection and feature extraction to enhance the consistency between the current frame features and the tracklet features, thus preventing identity swaps and tracklet breaks caused by object detection loss or distortion. Finally, a data association module based on hypergraph matching is constructed, integrating with object detection and feature extraction into a unified network, transforming the data association problem into a hypergraph matching problem between the tracklet hypergraph and the detection hypergraph, thereby achieving end-to-end model optimization. The experimental results demonstrate that this method has yielded favorable qualitative and quantitative analysis results on three multi-object tracking datasets, thereby validating its effectiveness in enhancing the robustness and accuracy of multi-object tracking tasks.

Motion blur is a common problem in the field of surveillance scenarios, and it obstructs the acquisition of valuable information. Thanks to the success of deep learning, a sequence of CNN-based architecture has been designed for image deblurring and has made great progress. As another type of neural network, transformers have exhibited powerful deep representation learning and impressive performance based on high-level vision tasks. Transformer-based networks leverage self-attention to capture the long-range dependencies in the data, yet the computational complexity is quadratic to the spatial resolution, which makes transformers infeasible for the restoration of high-resolution images. In this article, we propose an efficient transformer-based deblurring network, named MIMO-Uformer, for vehicle-surveillance scenarios. The distinct feature of the MIMO-Uformer is that the basic-window-based multi-head self-attention (W-MSA) of the Swin transformer is employed to reduce the computational complexity and then incorporated into a multi-input and multi-output U-shaped network (MIMO-UNet). The performance can benefit from the operation of multi-scale images by MIMO-UNet. However, most deblurring networks are designed for global blur, while local blur is more common under vehicle-surveillance scenarios since the motion blur is primarily caused by local moving vehicles. Based on this observation, we further propose an Intersection over Patch (IoP) factor and a supervised morphological loss to improve the performance based on local blur. Extensive experiments on a public and a self-established dataset are carried out to verify the effectiveness. As a result, the deblurring behavior based on PSNR is improved at least 0.21 dB based on GOPRO and 0.74 dB based on the self-established datasets compared to the existing benchmarks.

To address the challenges of change detection tasks, including the scarcity and dispersion of labeled samples, the difficulty in efficiently extracting features from unstructured image objects, and the underutilization of high-order correlation information, we propose a novel architecture based on hypergraph convolutional neural networks. By characterizing superpixel vertices and their high-order correlations, the method implicitly expands the number of labels while assigning adaptive weight parameters to adjacent objects. It not only describes changes in vertex features but also uncovers local and consistent changes within hyperedges. Specifically, a vertex aggregation mechanism based on superpixel segmentation is established, which segments the difference map into superpixels of diverse shapes and boundaries, and extracts their significant statistical features. Subsequently, a dynamic hypergraph structure is constructed, with each superpixel serving as a vertex. Based on the multi-head self-attention mechanism, the connection probability between vertices and hyperedges is calculated through learnable parameters, and the hyperedges are generated through threshold filtering. Moreover, a framework based on hypergraph convolutional neural networks is customized, which models the high-order correlations within the data through the learning optimization of the hypergraph, achieving change detection in remote sensing images. The experimental results demonstrate that the method obtains impressive qualitative and quantitative analysis results on the three remote sensing datasets, thereby verifying its effectiveness in enhancing the robustness and accuracy of change detection.

In vitro detection of small molecules in the human circulatory system contributes a lot to prompt personalized healthcare and early disease diagnose. The precise detection requires to select target molecules for real‐time noninvasive identifications, which remains a challenge derived from the lack of both molecule‐sieving techniques and rapid detection tools. Herein, positively charged polymer chains into graphene oxide membranes are seeded to tune their interlayer spacing precisely and persistently within a range of 8.67–13.75 Å. The adaptive yet stable nanocapillaries allow membranes to sieve small molecules with fitted sizes, for example, salts and biomarkers. The molecule‐sieving membranes with organic electrochemical transistors, further achieving an in vitro detection of lactate and uric acid with a high accuracy, are coupled. Based on this finding, a wearable sweat lactate rapid test kit, which detects sweat lactate secreted from human body within a few minutes, is demonstrated.

Continuous monitoring of lower extremity muscles is necessary, as the muscles support many human daily activities, such as maintaining balance, standing, walking, running, and jumping. However, conventional electromyography and physiological cross-sectional area methods inherently encounter obstacles when acquiring precise and real-time data pertaining to human bodies, with a notable lack of consideration for user comfort. Benefitting from the fast development of various fabric-based sensors, this paper addresses these current issues by designing an integrated smart compression stocking system, which includes compression garments, fabric-embedded capacitive pressure sensors, an edge control unit, a user mobile application, and cloud backend. The pipeline architecture design and component selection are discussed in detail to illustrate a comprehensive user-centered STIMES design. Twelve healthy young individuals were recruited for clinical experiments to perform maximum voluntary isometric ankle plantarflexion contractions. All data were simultaneously collected through the integrated smart compression stocking system and a muscle force measurement system (Humac NORM, software version HUMAC2015). The obtained correlation coefficients above 0.92 indicated high linear relationships between the muscle torque and the proposed system readout. Two-way ANOVA analysis further stressed that different ankle angles (p = 0.055) had more important effects on the results than different subjects (p = 0.290). Hence, the integrated smart compression stocking system can be used to monitor the muscle force of the lower extremities in isometric mode.

Accurately estimating channel state information is essential for meeting the quality-of-service requirements of modern applications and scenarios. Deep learning techniques have proven effective in acquiring channel conditions with low pilot overhead in massive connectivity scenarios. However, accessing channel data brings new challenges related to transmission overhead, privacy concerns, scalability, heterogeneous network support, and adaptability to dynamic environments. We propose a federated generative-adversarial-network-enabled channel estimator to address these challenges. We refine the coarse least-squares estimation results for their low complexity and fast convergence. To ensure accuracy, we designed a double U-shaped network. The Lipschitz continuous function is applied to discriminators for spectral normalization. We then propose a federated learning framework to utilize the training process. The local generator parameters are updated at the center, reducing communication overhead and privacy concerns. To deal with nonindependent and identically distributed datasets, the discriminators dynamically push away the predictions by dynamic regularization to obtain a more robust aggregated generative model at the center. Furthermore, we propose a motivation scheme that benefits users participating in the training process, encouraging them to join and take advantage of edge/cloud computing capabilities. Numerical results demonstrate that the proposed federated generative adversarial network-enabled channel estimator provides high estimation accuracy and reduces the burden on pilots. The proposed dynamic regularization terms and motivation scheme boost performance efficiently with low communication cost and high participation.

With the cross-integration of artificial intelligence and communications, technologies for processing multimodal data such as text, image, audio, and video are booming, the shared dimension of modal semantics is deeply excavated, and the characteristics of multimodal semantic information such as high abstraction, intelligence and simplicity are being fully utilized, which brings new ideas and means to semantic communications.First, the fundamental theories and classifications of semantic communication were introduced, and the research status of single-modal semantic communication was reviewed for text, image, audio, and video respectively.Then, the research status of multimodal semantic communication was reviewed, and multimodal data fusion technology and secure semantic communication were introduced.Finally, the challenges faced by multimodal semantic communication were summarized.

Abstract The main obstacle of ionic thermo‐electrochemical cells (TECs) in continuous power supply lies in a low heat‐to‐electricity energy conversion efficiency because most TECs work in thermodiffusion mode in which the ions are confined in a liquid/electrolyte media. The introduction of the redox couple onto the electrode surface may overcome the obstacle by resolving the low mass transport rate of ions caused by the redox process occurring near but not on the electrode surface. Herein, the authors demonstrate enhancement of TECs by integrating the redox couple directly onto the electrode surface to maximize the mass transport efficiency. A discontinuous interfacial modification strategy is developed by using a carbon cloth/iron (II/III) phytate as the symmetric electrodes. The gelled electrolyte consisting of a polyacrylamide matrix and phytic acid is shown to promote selective ion diffusion. A synergistic combination consisting of the thermodiffusion effect and redox reactions on the electrode is established in a pre‐treated layout. Such TEC affords a high output voltage of 0.4 V, an excellent instantaneous output power density (20.26 mW m‐2 K‐2) and a record‐high 2 h output energy density (2451 J m‐2) under TH = 30 °C with TC = 15 °C, with an ultrahigh Carnot‐relative efficiency of 1.12%.

Light and flexible thermoelectric generators operating at room temperature are highly desirable for wearable microelectronics. Here, flexible thermoelectric composites comprising semiconducting Bi2Te3 particles and conductive polymers exhibit a high output power within a small temperature window around room temperature.

Strong, electroactive and ultrafine conducting polymer fibres are prepared at the large scale by a modified wet spinning strategy. The obtained ultrafine polyaniline fibres are qualified to build tactile organic electrochemical transistors.

In the upcoming sixth-generation (6G) era, the demand for constructing a wide-area time-sensitive Internet of Things (IoT) continues to increase. As conventional cellular technologies are difficult to directly use for wide-area time-sensitive IoT, it is beneficial to use non-terrestrial infrastructures, including satellites and unmanned aerial vehicles (UAVs). Thus, we can build a non-terrestrial network (NTN) using a cell-free architecture. Driven by the time-sensitive requirements and uneven distribution of IoT devices, the NTN must be empowered using mobile edge computing (MEC) while providing oasis-oriented on-demand coverage for devices. Nevertheless, communication and MEC systems are coupled with each other under the influence of a complex propagation environment in the MEC-empowered NTN, which makes it difficult to coordinate the resources. In this study, we propose a process-oriented framework to design communication and MEC systems in a time-division manner. In this framework, large-scale channel state information (CSI) is used to characterize the complex propagation environment at an affordable cost, where a nonconvex latency minimization problem is formulated. Subsequently, the approximated problem is provided, and it can be decomposed into sub-problems. These sub-problems are then solved iteratively. The simulation results demonstrated the superiority of the proposed process-oriented scheme over other algorithms, implied that the payload deployments of UAVs should be appropriately predesigned to improve the efficiency of using resources, and confirmed that it is advantageous to integrate NTN with MEC for wide-area time-sensitive IoT.

Flexible electroadhesive clutches with high shear stress and fast response working at low voltage are much desired in wearable electronics and robotic systems. Dielectric materials with opposite charge characteristics could maximize the clutch performance by taking advantages of the boosted electroadhesion between the two contact pads in an asymmetrically structured clutch. In this paper, asymmetrically structured electroadhesive clutches are proposed and reported for the first time. The asymmetric structured clutch exhibits a two-fold increment in the shear force but similar response time by simply reversing the electrode polarity. This work provides a new dimension to realize high-performance electroadhesive clutches based on an asymmetric strategy.

Abstract Nowadays, the need for protective devices at man–machine interfaces is increasing in the fields of traffic, sports, construction, and military, etc. Dynamic pressure sensing technology with wide measuring range, high sensitivity, softness, and fast response is crucial for evaluation and optimization of the personal protective equipment under impact scenarios. However, current sensors hardly possess all the aforesaid required characteristics. For the first time, this article reports the evaluation and application of an innovative soft pressure sensor with modulus of 2 MPa, maximum pressure of 8 MPa, and over 500‐Hz frequency. A theoretical model, taking strain rate into consideration, is established to characterize the dynamic sensing behavior. A sensing network in the form of smart clothing is developed and used in a sled crash test, which is a standard approach to evaluate the safety of automobiles in collisions. The pressure distribution over the dummy's surface during the crash is acquired in real‐time, and compared with numerical simulations. This study is important to the study of occupant injury and crashworthiness design for vehicles, and it will benefit the automotive industry. With the built‐in sensing network, the smart clothing has promising applications in the pressure mapping of 3D flexible man–machine interface under impact scenarios.

AimGastrointestinal discomfort is the most common adverse event in metformin treatment for type 2 diabetes. The mechanism of action of metformin is associated with gut microbiota. However, the gut microbial community structure related to metformin-induced gastrointestinal adverse events remains unclear. This study aimed to investigate it.Methods50 patients with newly diagnosed diabetes were treated with metformin 1500mg/d for 12 weeks. The patients were divided into two groups according to whether gastrointestinal adverse events occurred (group B) or did not occur (group A) after treatment. The fecal bacterial communities and short-chain fatty acids (SCFAs) were sequenced and compared. 70 diabetes mice were randomly divided into 8 groups and treated with metformin (Met), clindamycin (Clin) and/or SCFA, which were the Met+/Clin+, Met+/Clin-, Met-/Clin+, Met-/Clin-, Met+/SCFA+, Met+/SCFA-, Met-/SCFA+ and Met-/SCFA- group. After 4 weeks of metformin treatment, blood glucose, food intake, fecal SCFAs, gut microbiota and gut hormones were measured.ResultsMetformin increased the abundance of Phascolarctobacterium, Intestinimonas and Clostridium III. Functional prediction analysis showed that the propanoate metabolism pathway was significantly up-regulated. The concentrations of acetic acid and propanoic acid in feces were significantly increased. The abundance of Clostridium sensu stricto, Streptococcus and Akkermansia induced by metformin in group B was higher than that in group A. The propanoate metabolism pathway and propanoic acid in feces were significantly up-regulated in group B. In the animal experiments, the food intake decreased and glucose control increased in metformin groups compared with those in the control groups. The total GLP-1 level in the Met+/Clin- group was significantly higher than that in the Met-/Clin- group, while there was no statistical difference between the Met-/Clin- and Met+/Clin+ group. The total GLP-1 level in the Met-/SCFA+ group was significantly higher than that in the Met-/SCFA-group, while the levels of total GLP-1 and active GLP-1 in the Met+/SCFA- group and the Met+/SCFA+ group were significantly higher than those in the Met-/SCFA-group.ConclusionsOur data suggest that metformin promotes the secretion of intestinal hormones such as GLP-1 by increasing the abundance of SCFA-producing bacteria, which not only plays an anti-diabetic role, but also may causes gastrointestinal adverse events.

The manufacturing of flexible strain sensors for wearable electronics usually requires different conductive materials for the sensing part and the connection part. This increases the complexity, cost, and performance issues due to the mismatch of the thermo-electro-mechanical properties of the materials. Herein, a new design scheme using a single conductive material is presented for a low-cost mass-producible fabric strain sensor, where a carbon/silicone nanocomposite is screen-printed to make both parts. By exploring the dimension effect and modelling of the conductive tracks, and adopting a large difference of over 100 times in aspect ratio, this research makes the electrical response of the fabric strain sensor depend almost exclusively on the sensing part, while its connection part has a low resistance. The sensor exhibits outstanding performance with a wide working range (60% strain), adequate linearity, long fatigue life (∼50,000 cycles), and mechanical robustness, rendering it suitable for human body movement detection. Moreover, the manufacturing process is simple and low-cost ($11 per m2). Thus, the new design scheme overcomes the mismatch issue and provides an important reference value for the design of flexible resistive sensors working in a high resistance range, from ∼ 100 KΩ to several MΩ.

Data-driven methods have shown promising results in structural health monitoring (SHM) applications. However, most of these approaches rely on the ideal dataset assumption and do not account for missing data, which can significantly impact their real-world performance. Missing data is a frequently encountered issue in time series data, which hinders standardized data mining and downstream tasks such as damage identification and condition assessment. While imputation approaches based on spatiotemporal relations among monitoring data have been proposed to handle this issue, they do not provide additional helpful information for downstream tasks. This paper proposes a robust deep learning-based method that unifies missing data imputation and damage identification tasks into a single framework. The proposed approach is based on a long short-term memory (LSTM) structured autoencoder (AE) framework, and missing data is simulated using the dropout mechanism by randomly dropping the input channels. Reconstruction errors serve as the loss function and damage indicator. The proposed method is validated using the quasi-static response (cable tension) of a cable-stayed bridge released in the 1st IPC-SHM, and results show that missing data imputation and damage identification can be effectively integrated into the proposed unified framework.

Previously reported wearable systems for people with Parkinson’s disease (PD) have been focused on the detection of abnormal gait. They suffered from limited accuracy, large latency, poor durability, comfort, and convenience for daily use. Herewith we report an intelligent wearable system (IWS) that can accurately detect abnormal gait in real-time and provide timely cueing for PD patients. The system features novel sensitive, comfortable and durable plantar pressure sensing insoles with a highly compressed data set, an accurate and fast gait algorithm, and wirelessly controlled timely sensory cueing devices. A total of 29 PD patients participated in the first phase without cueing for developing processes of the algorithm, which achieved an accuracy of over 97% for off-line detection of freezing of gait (FoG). In the second phase with cueing, the evaluation of the whole system was conducted with 16 PD subjects via trial and a questionnaire survey. This system demonstrated an accuracy of 94% for real-time detection of FoG and a mean latency of 0.37 s between the onset of FoG and cueing activation. In questionnaire survey, 88% of the PD participants confirmed that this wearable system could effectively enhance walking, 81% thought that the system was comfortable and convenient, and 70% overcame the FoG. Therefore, the IWS makes it an effective, powerful, and convenient tool for enhancing the mobility of people with PD.

Background. The Somogyi effect is defined as fasting hyperglycemia secondary to nocturnal hypoglycemia. In past decades, this effect proved to be rare or absent. However, many endocrinologists still believe in this phenomenon in clinical practice. Does the Somogyi effect truly exist? We aimed to answer this question with a study based on a larger sample size. Methods. We collected retrospective CGMs data from 2,600 patients with type 2 diabetes with stable treatment of insulin. Nocturnal hypoglycemia was defined as a CGMs sensor glucose of less than 3.9 mmol/L for at least 15 min between 24:00 and 06:00. Morning fasting glucose was compared between people with nocturnal hypoglycemia and without nocturnal hypoglycemia. Results. Valid CGMs data were obtained on 4,705 of 5,200 nights. Morning fasting glucose was observed lower after nights with nocturnal hypoglycemia compared with nights without hypoglycemia (P

Textile electronics have become an indispensable part of wearable applications because of their large flexibility, light-weight, comfort and electronic functionality upon the merge of textiles and microelectronics. As a result, the fabrication of functional fibrous materials and the integration of textile electronic devices have attracted increasing interest in the wearable electronic community. Challenges are encountered in the development of textile electronics in a way that is electrically reliable and durable, without compromising on the deformability and comfort of a garment, including processing multiple materials with great mismatches in mechanical, thermal, and electrical properties and assembling various structures with the disparity in dimensional scales and surface roughness. Equal challenges lie in high-quality and cost-effective processes facilitated by high-level digital technology enabled design and manufacturing methods. This work reviews the manufacturing of textile-shaped electronics via the processing of functional fibrous materials from the perspective of hierarchical architectures, and discusses the heterogeneous integration of microelectronics into normal textiles upon the fabric circuit board and adapted electrical connections, broadly covering both conventional and advanced textile electronic production processes. We summarize the applications and obstacles of textile electronics explored so far in sensors, actuators, thermal management, energy fields, and displays. Finally, the main conclusions and outlook are provided while the remaining challenges of the fabrication and application of textile electronics are emphasized.

Piezoelectric catalysis (piezocatalysis) is a physical/chemical process that utilizes piezoelectric potential for accelerating chemical reactions, in which ubiquitous mechanical energies in nature are used for various catalysis applications, e.g., treating organic water pollutants. Despite the high efficiency achieved by piezocatalytic powders, the particles used tend to diffuse in water systems and are hard to be separated, thus causing secondary pollution. Herein, a free‐standing piezocatalytic foam is designed and fabricated, which is composed of BaTiO3 nanoparticles embedded in the PVDF scaffold. The as‐prepared PVDF–BaTiO3 composite foam demonstrates outstanding piezocatalytic efficiency in removing aqueous organics among state‐of‐the‐art integral piezocatalytic platforms, which lie in the synergy of piezoelectric materials and abundant interconnected pores within the foam. Significantly, PVDF–BaTiO3 foam is further applied for purifying natural water samples, by which the permanganate index of the water sample reduces by nearly 30% after 2 h of treatment. In addition, as a monolithic platform, PVDF–BaTiO3 foam is easy to be collected, with high reuse stability and applicability for treating various pollutants, resulting in dominant advantages over powder‐based systems for practical high‐flux wastewater treatment. Herein, a piezocatalytic platform is provided for the effective degradation of organic pollutants in water, with minimal environmental side effects.

Abstract Solar‐driven overall water splitting based on metal sulfide semiconductor photocatalysts remains as a challenge owing to the strong charge recombination and deficient catalytic active sites. Additionally, significant inhibition of back reactions, especially the oxidation of sulfide ions during the photocatalytic water oxidation catalysis, is an arduous task that requires an efficient photogenerated hole transfer dynamics. Here, a ternary dumbbell‐shaped catalyst based on RuO2/CdS/MoS2 with spatially separated catalytic sites is developed to achieve simultaneous production of hydrogen and oxygen under simulated solar‐light without any sacrificial agents. Particularly, MoS2 nanosheets anchored on the two ends of CdS nanowires are identified as a reduction cocatalyst to accelerate hydrogen evolution, while RuO2 nanoparticles as an oxidation cocatalyst are deposited onto the sidewalls of CdS nanowires to facilitate oxygen evolution kinetics. The density functional theory simulations and ultrafast spectroscopic results reveal that photogenerated electrons and holes directionally migrate to MoS2 and RuO2 catalytic sites, respectively, thus achieving efficient charge carrier separation. The design of ternary dumbbell structure guarantees metal sulfides against photocorrosion and thus extends their range in solar water splitting.

Recent studies have greatly promoted the development of semantic segmentation. Most state-of-the-art methods adopt fully convolutional networks (FCNs) to accomplish this task, in which the fully connected layer is replaced with the convolution layer for dense prediction. However, standard convolution has limited ability in maintaining continuity between predicted labels as well as forcing local smooth. In this paper, we propose the dense convolution unit (DCU), which is more suitable for pixel-wise classification. The DCU adopts dense prediction instead of the center-prediction manner used in current convolution layers. The semantic label for every pixel is inferred from those overlapped center/off-center predictions from the perspective of probability. It helps to aggregate contexts and embeds connections between predictions, thus successfully generating accurate segmentation maps. DCU serves as the classification layer and is a better option than standard convolution in FCNs. This technique is applicable and beneficial to FCN-based state-of-the-art methods and works well in generating segmentation results. Ablation experiments on benchmark datasets validate the effectiveness and generalization ability of the proposed approach in semantic segmentation tasks.

In this paper, the geometry-aware GAN, referred to as GAGAN, is proposed to address the issue of face attribute transfer with unpaired data. The source and target images are not aligned and come from different individuals. The key idea is to deform the source image according to the geometry features and generate a high-resolution image with the desired attribute. To address the problem of the unpaired training samples, the CycleGAN architecture is applied to form an attribute adding and removing cycle, where the bilateral mappings between the source and target domains are learned. The geometry flow and occlusion mask are learned by the warping sub-network to capture the geometric variation between the two domains. In the attribute adding process, the spatial transformer network (STN) warps the source face into the desired pose and shape according to the flow, and the transfer sub-network hallucinates new components on the warped image. In the attribute removing process, the recover sub-network and the STN reverts the sample back to the source domain. Experiments on the benchmarks CELEBA and CELEBA-HQ datasets demonstrate the advantages of our method compared to the baselines, in terms of both quantitative and qualitative evaluation.

With the popularity of handheld devices, the development of wireless communication technology and the proliferation of multimedia resources, mobile video has become the main business in LTE networks with explosive traffic demands. How to improve the quality of experience (QoE) of mobile video in the dynamic and complex network environment has become a research focus. Dynamic adaptive streaming over HTTP technology introduces adaptive bitrate (ABR) requests at the client side to improve video QoE and various rate adaptation algorithms are also constantly proposed. In view of the limitations of the existing heuristic or learning-based ABR methods, we propose redirecting enhanced Deep Q-learning toward DASH video QoE (RDQ), a QoE-oriented rate adaptation framework based on enhanced deep Q-learning. First, we establish a chunkwise subjective QoE model and utilize it as the reward function in reinforcement learning so that the strategy can converge toward the direction of maximizing the subjective QoE score. Then, we apply several effective improvements of deep Q-learning to the RDQ agent’s neural network architecture and learning mechanism to achieve faster convergence and higher average reward than other learning-based methods. The proposed RDQ agent has been thoroughly evaluated using trace-based simulation on the real-time LTE network data. For disparate network scenarios and different video contents, the RDQ agent can outperform the existing methods in terms of the QoE score. The breakdown analysis shows that RDQ can suppress the number and the duration of the stalling events to the minimum while maintaining high video bitrate, thus achieving better QoE performance than other methods.

Quality of experience (QoE), which directly relates to both technical evolvement and profit promotion, is a vital concern for mobile video services. However, the wireless network operators have long been troubled by the problem of lacking effective QoE monitoring approaches since the traditional evaluation methods of communication quality are objective metrics oriented. Considering that mobile terminals are the network elements closest to users, it is promising to realize a real-time QoE estimation for video services by fully utilizing the sensing capabilities of mobile terminals. As the first step, we specifically develop a mobile video testing application. With support from China Unicom, one of the three major wireless network operators in China, over 80 000 data records are collected under the real-world conditions. The collected data consist of four types of subjective scores and 13 objective parameters concerning video attributes, network performance, device capability, playback events, and external factors. After preprocessing the data set through correlation analysis, we establish the two QoE estimation models based on the C4.5 method and the gradient boosting decision tree (GBDT) method, respectively. The experimental results demonstrate that the proposed models can achieve remarkable estimation performances and outperform the baseline models. Specifically, the overall estimation accuracy of the GBDT-based model is approximately 80% for a five-level scale and approaches 90% when a more practical 3-level scale is adopted. Finally, we comprehensively discuss the estimation performances based on characteristics of the data and validate the feasibility of estimating QoE based on mobile terminals-the “peripheral sensors” of the mobile networks.

Flexible sensors based on the highly elastic Polyurethane (PU)-containing textile substrates have been frequently incorporated in enormous wearable applications. However, the desirable sensing properties such as stable sensitivity, small hysteresis and good repeatability depend on the mechanical resilience of the textile substrates. This paper conducts a systematic investigation into the mechanical hysteresis of the PU-containing textile substrates, from fibers to yarns and then fabrics. The impact of fiber materials, yarn structures, and fabric constructions on the overall tensile properties of the PU-containing substrates were examined. It was found that fibers of high elasticity, stable yarn structure with excellent recovery, as well as appropriate fabric construction are effective to reduce the mechanical hysteresis of the textile substrates. Coated strain sensors based on different substrates were then fabricated. Results show that smaller mechanical hysteresis of the PU-containing substrates generally led to lower electromechanical hysteresis and better repeatability of sensors. This paper gives all-round consideration of the fibers and structure features of substrates to provide recommendation for reducing mechanical and electromechanical hysteresis of sensors based on PU-containing textiles and benefiting the design and optimization of other flexible sensors. Keywords: Polyurethane-containing textile, Textile substrates, Mechanical hysteresis, Coated sensors, Electromechanical hysteresis

Transduction efficiency in electrochemical actuators hardly exceeds 1% because the current electrode materials do not allow to efficiently exploit microstructural changes to achieve large actuation effects. Here the authors demonstrate transduction efficiencies up to 6% in actuators by using graphdiyne as electrode material.

This paper attempts to fuse a multispectral image and an auxiliary hyperspectral image (HSI) with no requirement of image registration. Most previous studies solve this problem with sparsity-based methods. However, in this paper, a novel fusion framework is developed based on a Gaussian mixture model (GMM): First, the GMM is adopted to extract the spectral information from the input HSI. Low-rank constraints are imposed on the covariance matrices of the model to solve the computational problem in the expectation-maximization approach. Second, considering the spatial self-similarity, a structure-similarity regularization term is designed to further enhance the quality of the reconstructed image. To that end, a forward–backward splitting method is adopted to cut down the computational complexity of the optimization. The proposed method does not require two well-aligned images, thus, it will not be influenced by the registration errors between two fusing images. Experimental results of a simulated data set and an actual satellite (EO-1/Hyperion/ALI) data set show that the proposed method displays a stable performance and outperforms many state-of-the-art methods with acceptable computational complexity, when registration errors are taken into consideration.

This paper presents a hierarchical multinomial latent model with G0 distribution for synthetic aperture radar (SAR) image semantic segmentation. The model considers the scattering statistics and multiscale characteristics of the SAR images. Specifically, a 2-D discrete wavelet transform is used to construct a quad-tree structure, and a hierarchical multinomial latent model is used for semantic segmentation. The hierarchical model can capture local and global characteristics of the SAR images. Additionally, statistical distributions of SAR images are discussed. The flexible G0 distribution is substituted into the hierarchical multinomial latent model for the segmentation of various types of land covers. Moreover, the developed Bayesian inference on the quad-tree structure is incorporated into our approach. The semantic segmentation map is obtained from an initialization, bottom-up, and top-down probability computations. In this way, the underlying spatial structure information is effectively propagated. Our experiments compare the proposed approach, other multinomial latent model-based methods, and Markov random field-based methods. The experiments are conducted from a series of synthetic and real SAR images, where the segmentation results demonstrate that our approach is robust to the noise in most cases, obtains the best result among the compared methods and improves the state-of-the-art segmentation performance.

Abstract A new bio-mechanical model for elbow flexions is proposed to quantify the elbow torque generated as a function of the upper-arm circumferential strain and influencing factors of elbow angle and angular velocity. The upper-arm circumferential strain is used to represent the contractile intensity of the dominant flexor, biceps brachii, whose behavior is described by Hill’s theory. Experiments with thirteen healthy subjects were conducted to determine the influencing factors. The temporal distributions of torque and elbow angle were measured by Biodex ®3 simultaneously, while the upper-arm circumference was obtained by a wearable anthropometric measurement device. Within the experimental range, the change of angular velocity has been found to have no effect on the torque generated. The new model was further verified experimentally with reasonable agreements obtained. The mean relative error of the torque estimated from the model is 15% and 22%, for isokinetic and isotonic flexions, respectively. The verified model establishes the relationship between the torque generated and circumference strain of the upper arm, for the first time, thus provide a scientific foundation for the anthropometric measurement technology as an alternative to sEMG for monitoring force/torque generation during elbow flexions.

Abstract Breath figures (BFs) are a kind of water droplet arrays that can be formed by condensing aqueous vapor onto a cold surface, such as dewy phenomenon on a spider web. This study developed a BF-inspired approach for direct introduction of desired materials onto the textile surfaces with three-dimensionally conformal porous microstructures by the evaporation of solution-coated fabric under high humidity environment, which brings a brand-new kind of modified textiles, three-dimensionally conformal porous microstructured fabrics (CPMFs). Such kind of CPMFs can possess customized multifunctional properties of introduced materials, and meanwhile maintain the inherent properties and unique texture features of fabrics. This nature-inspired BF approach is robust and versatile for customized preparation of CPMFs based on different fabrics with different common polymers. Moreover, it is also feasible for one-step functionalization of CPMFs by the incorporation of nanoparticles (such as titanium dioxide nanoparticles, TiO2 NPs) into the porous microstructures during the BF process. Comparing to the sample modified without porous microstructures, the resultant TiO2 NPs-incorporated CPMFs show an obviously enhanced performance on photocatalytic degradation of pollutants under the same ultraviolet irradiation conditions.

In the fields of humanoid robots, soft robotics, and wearable electronics, the development of artificial skins entails pressure sensors that are low in modulus, high in sensitivity, and minimal in hysteresis. However, few sensors in the literature can meet all the three requirements, especially in the low pressure range (−1) to compression forces below 0.65 N (6.8 kPa). The low-end detection limit is ~0.0012 N (13 Pa), only triple the mass of a bee. Minimal resistance hysteresis of the pressure sensor is 7.7%. The low hysteresis is attributed to the study on the carbon/silicone nanocomposite, which reveals the effect of heat treatment on its mechanical and electromechanical hysteresis. Pressure measurement range and sensitivity of the sensor can be tuned by changing the structure and strain gauge parameters. This concept of sensor design, when combined with microfluidics technology, is expected to enable soft, stretchable, and highly precise touch-sensitive artificial skins.

The continuous development of electron devices towards the trend of “More than Moore” requires functional diversification that can collect data (sensors) and store (memories) and process (computing units) information. Considering the large occupation proportion of image data in both data center and edge devices, a device integration with optical sensing and data storage and processing is highly demanded for future energy-efficient and miniaturized electronic system. Two-dimensional (2D) materials and their heterostructures have exhibited broadband photoresponse and high photoresponsivity in the configuration of optical sensors and showed fast switching speed, multi-bit data storage, and large ON/OFF ratio in memory devices. In addition, its ultrathin body thickness and transfer process at low temperature allow 2D materials to be heterogeneously integrated with other existing materials system. In this paper, we overview the state-of-the-art optoelectronic random-access memories (ORAMs) based on 2D materials, as well as ORAM synaptic devices and their applications in neural network and image processing. The ORAM devices potentially enable direct storage/processing of sensory data from external environment. We also provide perspectives on possible directions of other neuromorphic sensor design (e.g., auditory and olfactory) based on 2D materials towards the future smart electronic systems for artificial intelligence.

This paper presents a review of the development of optical fibers made of multiple materials, particularly including silica glass, soft glass, polymers, hydrogels, biomaterials, Polydimethylsiloxane (PDMS), and Polyperfluoro-Butenylvinyleth (CYTOP). The properties of the materials are discussed according to their various applications. Typical fabrication techniques for specialty optical fibers based on these materials are introduced, which are mainly focused on extrusion, drilling, and stacking methods depending on the materials’ thermal properties. Microstructures render multiple functions of optical fibers and bring more flexibility in fiber design and device fabrication. In particular, micro-structured optical fibers made from different types of materials are reviewed. The sensing capability of optical fibers enables smart monitoring. Widely used techniques to develop fiber sensors, i.e., fiber Bragg grating and interferometry, are discussed in terms of sensing principles and fabrication methods. Lastly, sensing applications in oil/gas, optofluidics, and particularly healthcare monitoring using specialty optical fibers are demonstrated. In comparison with conventional silica-glass single-mode fiber, state-of-the-art specialty optical fibers provide promising prospects in sensing applications due to flexible choices in materials and microstructures.

Precise measurements of low pressure are highly necessary for many applications. This study developed novel structured fibre sensors embedded in silicone, forming smart skin with high sensitivity, high durability, and good immunity to crosstalk for precise measurement of pressure below 10 kPa. The transduction principle is that an applied pressure leads to bending and stretching of silicone and optical fibre over a purposely made groove and induces the axial strain in the gratings. The fabricated sensor showed high pressure sensitivity up to 26.8 pm/kPa and experienced over 1,000,000 cycles compression without obvious variation. A theoretical model of the sensor was presented and verified to have excellent agreement with experimental results. The prototype of smart leg mannequin and wrist pulse measurements indicated that such optical sensors can precisely measure low-pressure and can easily be integrated for smart skins for mapping low pressure on three-dimensional surfaces.

Aims. We investigated whether poor sleep quality is associated with both dawn phenomenon and impaired circadian clock gene expression in subjects with diabetes. Methods. 81 subjects with diabetes on continuous glucose monitoring were divided into two groups according to the Pittsburgh Sleep Quality Index. The magnitude of dawn phenomenon was quantified by its increment from nocturnal nadir to prebreakfast. Peripheral leucocytes were sampled from 81 subjects with diabetes and 28 normal controls at 09:00. Transcript levels of circadian clock genes (BMAL1, PER1, PER2, and PER3) were determined by real-time quantitative polymerase chain reaction. Results. The levels of HbA1c and fasting glucose and the magnitude of dawn phenomenon were significantly higher in the diabetes group with poor sleep quality than that with good sleep quality. Peripheral leucocytes from subjects with poor sleep quality expressed significantly lower transcript levels of BMAL1 and PER1 compared with those with good sleep quality. Poor sleep quality was significantly correlated with magnitude of dawn phenomenon. Multiple linear regression showed that sleep quality and PER1 were significantly independently correlated with dawn phenomenon. Conclusions. Dawn phenomenon is associated with sleep quality. Furthermore, mRNA expression of circadian clock genes is dampened in peripheral leucocytes of subjects with poor sleep quality.