Search Results (939)

Transformers have emerged as the central backbone architecture for modern generative AI. However, most ML applications targeting low-power, low-cost SoCs (TinyML apps) do not employ Transformers as these models are thought to be challenging to quantize and deploy on small devices. This work proposes a methodology to reduce Transformer dimensions with an extensive pruning search. We exploit the intrinsic redundancy of these models to fit them on resource-constrained devices with a well-controlled accuracy tradeoff. We then propose an optimized library to deploy the reduced models using BFLoat16 with no accuracy loss on Commercial Off-The-Shelf (COTS) RISC-V multi-core micro-controllers, enabling the execution of these models at the extreme edge, without the need for complex and accuracy-critical quantization schemes. Our solution achieves up to 220× speedup with respect to a naïve C port of the Multi-Head Self Attention PyTorch kernel: we reduced MobileBert and TinyViT memory footprint up to ∼94% and ∼57%, respectively, and we deployed a tinyLLAMA SLM on microcontroller, achieving a throughput of 1219 tokens/s with an average power of just 57 mW. Full article

Research on triboelectric nanogenerators (TENGs) and self-powered devices has rapidly grown in recent years since its first report in 2012 by Prof. Wang’s group. Triboelectric polymers have been a frontier of the research, attributed to their high surface potential and consequently high voltage output. To further advance the field, in recent years, photoactive semiconductor materials have been introduced which offer an additional current generation mechanism under light excitation, boosting the output current of the TENG. In addition, the semiconductor-based TENG further provides an ability to detect photo-signals beyond mechanical signals, adding high value towards advanced multi-functional sensor applications. In this regard, this article aims to review the recent progress in semiconductor-based TENGs, particularly on metal-halide perovskites, and their applications to self-powered electronics. Finally, the prospects and challenges of the perovskite-based TENG are discussed. Full article

The progression of wearable technology has revealed that cellulose-based triboelectric nanogenerators (TENG) possess considerable promise in self-powered micro-sensing technology; this is attributed to their superior biocompatibility, sustainability, and mechanical characteristics. This paper aims to explore the application of the cellulose-based TENG self-powered micro-sensing technology in wearable systems for human health monitoring. First, the working principles and modes of TENG are summarized, along with the characteristics of the cellulose, nanocellulose, cellulose derivatives and the advantages of the cellulose-based TENG. Next, we discuss in detail the applications of the cellulose-based TENG in monitoring physiological parameters, such as heart rate, motion, respiration, and pulse, and we analyze their advantages and challenges in practical applications. Additionally, we explore the integration of the cellulose-based TENG human–machine interaction sensors in health monitoring devices. Finally, we outline the current challenges and future research directions in this field, including the enhancement of triboelectric performance, adaptability to diverse environments, controllable degradability, and multi-scenario real-world applications. This review provides a comprehensive perspective on the application of the cellulose-based TENG self-powered micro-sensing technology in wearable health monitoring systems and offers guidance for future research and development. Full article

Wearable electronic devices have shown great application prospects in the fields of tactile sensing, electronic skin, and soft robots. However, the existing wearable electronic devices face limitations such as power supply challenges, lack of portability, and discomfort, which restrict their applications. The invention of triboelectric nanogenerators (TENGs) with dual functions of energy harvesting and sensing provides an innovative solution to address these issues. This study prepared a highly stretchable conductive hydrogel using doped conducting polymer poly(3,4-ethylenedioxythiophene):poly(styrene sulfonate) (PEDOT:PSS) as a strain sensor, demonstrating high sensitivity (GF = 4.31), an ultra-wide sensing range (0–1690%), ultra-fast response speed (0.15 s), excellent durability, and repeatability. A high-performance triboelectric nanogenerator was constructed using the hydrogel as an electrode, achieving an output performance of up to 192 V. Furthermore, the TENG fixed in the hands, wrists, legs, and feet of the human body can be used as a wearable electronic device to monitor human motion, which is conducive to promoting the development of triboelectric nanogenerators based on conductive hydrogels in strain sensors and self-powered wearable devices. Full article

Agricultural sustainability is becoming more and more important for human health. Wireless sensing technology could provide smart monitoring in real time for different parameters in planting, breeding, and the food supply chain with advanced sensors such as flexible sensors; wireless communication networks such as third-, fourth-, or fifth-generation (3G, 4G, or 5G) mobile communication technology networks; and artificial intelligence (AI) models. Many sustainable, natural, renewable, and recycled facility energies such as light, wind, water, heat, acoustic, radio frequency (RF), and microbe energies that exist in actual agricultural systems could be harvested by advanced self-powered technologies and devices using solar cells, electromagnetic generators (EMGs), thermoelectric generators (TEGs), piezoelectric generators (PZGs), triboelectric nanogenerators (TENGs), or microbial full cells (MFCs). Sustainable energy harvesting to the maximum extent possible could lead to the creation of sustainable self-powered wireless sensing devices, reduce carbon emissions, and result in the implementation of precision smart monitoring, management, and decision making for agricultural production. Therefore, this article suggests that proposing and developing a self-powered wireless sensing system for sustainable agriculture (SAS) would be an effective way to improve smart agriculture production efficiency while achieving green and sustainable agriculture and, finally, ensuring food quality and safety and human health. Full article

Energy harvesting technologies are becoming increasingly popular as potential sources of energy for Internet of Things (IoT) devices. Magnetic field energy harvesting (MFEH) from current-carrying components, such as power cables, represents a particularly promising technology for smart grid, infrastructure, and environmental monitoring applications. This paper presents a single-stage AC/DC power converter, a control architecture, and an energy harvester design applicable to MFEH devices. The power converter consists of a MOSFET full bridge that is used to actively rectify the induced voltage at the transceiver while providing a regulated output voltage. The approach is suitable for a broad range of grid power lines, offering a compact power stage that achieves a reduction in component count while active rectification minimizes energy losses, thereby improving thermal management in power electronics compared with the previous research. The experimental results demonstrate that the power converter provides a stable energy source and offers an alternative to self-powering smart grid IoT devices. Full article

Nowadays, the proliferation of distributed renewable energy sources is a fact. A microgrid is a good solution to self-manage the energy generation and consumption of electrical loads and sources from the point of view of the consumer as well as the power system operator. To make a microgrid as versatile as necessary to carry that out, a flexible inverter is necessary. In this paper, an algorithm is presented to control an inverter and make it complete and versatile to work in grid-connected and in isolated modes, injecting or receiving power from the grid and always compensating the harmonics generated by the loads in the microgrid. With this inverter, the microgrid can work while optimizing its energy consumption or according to the power system operator instructions. The inverter proposed is tested in a designed Matlab/Simulink simulation platform. After that, an experimental platform designed and built ad hoc, including a DC source, AC linear and non-linear loads, and a Semikron power inverter, is used to test the proposed control strategies. The results corroborate the good system performance. The replicability of the system is guaranteed by the use of low-cost devices in the implementation of the control. Full article

As an innovative branch of electronics, intelligent electronic textiles (e-textiles) have broad prospects in applications such as e-skin, human–computer interaction, and smart homes. However, it is still a challenge to distinguish multiple stimuli in the same e-textile. Herein, we propose a dual-parameter smart e-textile that can detect human pulse and body temperature in real time, with high performance and no signal interference. The doping of SWCNTs in PEDOT:PSS improves the electrical conductivity and Seebeck coefficient of the prepared composites, which results in excellent pressure and temperature-sensing properties of the PEDOT:PSS/SWCNTs/CS@PET-textile (PSCP) sensor. The dual-mode sensor has high sensitivity (32.4 kPa⁻¹), fast response time (~21 ms), and excellent durability (>2000 times) in pressure detection. Concurrently, this sensor maintains a high Seebeck coefficient of 25 μV/K in the 0–120 K temperature range with a tremendous linear relationship. Based on impressive dual-mode sensing characteristics and independent temperature-difference- and pressure-sensing mechanisms, smart e-textile sensors realize the real-time simultaneous monitoring of weak pulse signals and human body temperature, showing great potential in medical healthcare. In addition, the potential energy is excited by the temperature gradient between the human skin and the environment, which provides a novel idea for wearable self-powered devices. Full article

Tactile sensing is currently a research hotspot in the fields of intelligent perception and robotics. The method of converting external stimuli into electrical signals for sensing is a very effective strategy. Herein, we proposed a self-powered, flexible, transparent tactile sensor integrating sliding and proximity sensing (SFTTS). The principle of electrostatic induction and contact electrification is used to achieve tactile response when external objects approach and slide. Experiments show that the material type, speed, and pressure of the perceived object can cause the changes of the electrical signal. In addition, fluorinated ethylene propylene (FEP) is used as the contact electrification layer, and indium tin oxide (ITO) is used as the electrostatic induction electrode to achieve transparency and flexibility of the entire device. By utilizing the transparency characteristics of this sensor to integrate with optical cameras, it is possible to achieve integrated perception of tactile and visual senses. This has great advantages for applications in the field of intelligent perception and is expected to be integrated with different types of optical sensors in the future to achieve multimodal intelligent perception and sensing technology, which will contribute to the intelligence and integration of robot sensing. Full article

Hybrid organohalide perovskites have received considerable attention due to their exceptional photovoltaic (PV) conversion efficiencies in optoelectronic devices. In this study, we report the development of a highly sensitive, self-powered perovskite-based photovoltaic photodiode (PVPD) fabricated by incorporating a poly(amic acid)-polyimide (PAA-PI) copolymer as an interfacial layer between a methylammonium lead iodide (CH₃NH₃PbI₃, MAPbI₃) perovskite light-absorbing layer and a poly(3,4-ethylenedioxythiophene)-poly(styrene sulfonate) (PEDOT: PSS) hole injection layer. The PAA-PI interfacial layer effectively suppresses carrier recombination at the interfaces, resulting in a high power conversion efficiency (PCE) of 11.8% compared to 10.4% in reference devices without an interfacial layer. Moreover, applying the PAA-PI interfacial layer to the MAPbI₃ PVPD significantly improves the photodiode performance, increasing the specific detectivity by 49 times to 7.82 × 10¹⁰ Jones compared to the corresponding results of reference devices without an interfacial layer. The PAA-PI-passivated MAPbI₃ PVPD also exhibits a wide linear dynamic range of ~103 dB and fast response times, with rise and decay times of 61 and 18 µs, respectively. The improved dynamic response of the PAA-PI-passivated MAPbI₃ PVPD enables effective weak-light detection, highlighting the potential of advanced interfacial engineering with PAA-PI interfacial layers in the development of high-performance, self-powered perovskite photovoltaic photodetectors for a wide range of optoelectronic applications. Full article

In medium- and high-power-density applications, silicon carbide (SiC) metal-oxide semiconductor field effect transistors (MOSFETs) are often connected in parallel increasing the current capability. However, the current sharing of paralleled SiC MOSFETs is affected by the mismatched technical parameters of devices and the deviated power circuit parasitic inductances, even if power devices are controlled by a single gate driver. This leads to unevenly distributed power losses causing different stress between SiC MOSFETs. As a result, unbalanced current sharing increases the probability of severe power switch(es) and system failures. For over a decade, the current imbalance issue between parallel-connected SiC MOSFETs has concerned the scientific community, and many methods and techniques have been proposed. However, most of these solutions are impossible to realize without the necessity of screening power devices to measure their technical parameters. Consequently, system costs significantly increase due to the expensive equipment for screening SiC MOSFETs. Also, transient current imbalance is the main concern of most papers, without addressing static imbalance. In this paper, an innovative approach is proposed, capable of suppressing both static and transient current imbalance between paralleled SiC MOSFETs, under both symmetrical and asymmetrical layouts, through an improved active gate driver and without the requirement for any power device screening process. Additionally, the proposed solution employs a self-sustaining algorithmic approach utilizing current sensors and a field-programmable gate array (FPGA). The functionality of the proposed solution is verified through experimental tests, achieving current imbalance suppression between two paralleled SiC MOSFETs, actively and autonomously. Full article

Piezoelectric sensors convert mechanical stress into electrical charge via the piezoelectric effect, and when fabricated as fibers, they offer flexibility, lightweight properties, and adaptability to complex shapes for self-powered wearable sensors. Polyvinylidene fluoride (PVDF) nanofibers have garnered significant interest due to their potential applications in various fields, including sensors, actuators, and energy-harvesting devices. Achieving optimal piezoelectric properties in PVDF nanofibers requires the careful optimization of polarization. Applying a high electric field to PVDF chains can cause significant mechanical deformation due to electrostriction, leading to crack formation and fragmentation, particularly at the chain ends. Therefore, it is essential to explore methods for polarizing PVDF at the lowest possible voltage to prevent structural damage. In this study, a Design of Experiments (DoE) approach was employed to systematically optimize the polarization parameters using a definitive screening design. The main effects of the input parameters on piezoelectric properties were identified. Heat treatment and the electric field were significant factors affecting the sensor’s sensitivity and β-phase fraction. At the highest temperature of 120 °C and the maximum applied electric field of 3.5 kV/cm, the % β-phase (F(β)) exceeded 95%. However, when reducing the electric field to 1.5 kV/cm and 120 °C, the % F(β) ranged between 87.5% and 90%. The dielectric constant (ɛ′) of polarized PVDF was determined to be 30 at an electric field frequency of 1 Hz, compared to a value of 25 for non-polarized PVDF. The piezoelectric voltage coefficient (g₃₃) for polarized PVDF was measured at 32 mV·m/N at 1 Hz, whereas non-polarized PVDF exhibited a value of 3.4 mV·m/N. The findings indicate that, in addition to a high density of β-phase dipoles, the polarization of these dipoles significantly enhances the sensitivity of the PVDF nanofiber mat. Full article

This study evaluated the influence of cycle computers on the accuracy of power and cadence data. The research was divided into three phases: (1) a graded exercise test (GXT) at different constant loads to record power and cadence data; (2) a self-paced effort lasting 1 min to measure mean maximal power output (MMP); and (3) a short all-out effort. Eight cyclists completed the GXT, ten participated in the 1-min test, and thirty participated in the sprint effort. All participants pedaled on a controlled-resistance cycle ergometer, and the data were recorded using the ergometer itself and ten synchronized cycle computers of the same brand, configured to record at 1 Hz. The results showed minimal variations in power and cadence between devices during the GXT, suggesting adequate accuracy for constant efforts lasting a certain duration. However, in self-paced and high-intensity efforts (1-min and short all-out efforts), significant differences were observed between several devices, particularly in cadence and mean power, highlighting the relevance of device selection in these contexts. These findings suggest that, while variations in constant efforts may be negligible, in short-duration, high-intensity activities, the choice of device may be crucial for the accuracy and reliability of the data. Full article

As seabed exploration activities increase, side-scan sonar (SSS) is being used more widely. However, distortion and noise during the acoustic pulse’s travel through water can blur target details and cause feature loss in images, making target recognition more challenging. In this paper, we improve the YOLO model in two aspects: lightweight design and accuracy enhancement. The lightweight design is essential for reducing computational complexity and resource consumption, allowing the model to be more efficient on edge devices with limited processing power and storage. Thus, meeting our need to deploy SSS target detection algorithms on unmanned surface vessel (USV) for real-time target detection. Firstly, we replace the original complex convolutional method in the C2f module with a combination of partial convolution (PConv) and pointwise convolution (PWConv), reducing redundant computations and memory access while maintaining high accuracy. In addition, we add an adaptive scale spatial fusion (ASSF) module using 3D convolution to combine feature maps of different sizes, maximizing the extraction of invariant features across various scales. Finally, we use an improved multi-head self-attention (MHSA) mechanism in the detection head, replacing the original complex convolution structure, to enhance the model’s ability to focus on important features with low computational load. To validate the detection performance of the model, we conducted experiments on the combined side-scan sonar dataset (SSSD). The results show that our proposed SS-YOLO model achieves average accuracies of 92.4% (mAP 0.5) and 64.7% (mAP 0.5:0.95), outperforming the original YOLOv8 model by 4.4% and 3%, respectively. In terms of model complexity, the improved SS-YOLO model has 2.55 M of parameters and 6.4 G of FLOPs, significantly lower than those of the original YOLOv8 model and similar detection models. Full article

This study investigates the effects of negative bias temperature (NBT) stress and irradiation on the threshold voltage (V_T) of p-channel VDMOS transistors, focusing on degradation, recovery after each type of stress, and operational behavior under varying conditions. Shifts in V_T (ΔV_T) were analyzed under different stress orders, showing distinct influence mechanisms, including defects creation and their removal and electrochemical reactions. Recovery data after each type of stress indicated ongoing electrochemical processes, influencing subsequent stress responses. Although the ΔV_T is not particularly pronounced during the recovery after irradiation, changes in subthreshold characteristics indicate the changes in defect densities that affect the behavior of the components during further application. Additionally, the findings show that the ΔV_T during the NBT stress after irradiation (up to certain doses and conditions) remains relatively stable, but this is the result of a balance of competing mechanisms. A subthreshold characteristic analysis provided a further insight into the degradation dynamics. A particular attention was paid to analyzing ΔV_T with a focus on predicting the lifetime. In practical applications, especially under pulsed operation, prior stresses altered the device’s thermal and electrical performance. It was shown that self-heating effects were more pronounced in pre-stressed components, increasing the power dissipation and thermal instability. These insights additionally highlight the importance of understanding stress-induced degradation and recovery mechanisms for optimizing VDMOS transistor reliability in advanced electronic systems. Full article