Search

Featured Application: These results demonstrate that a BLE wireless sensor system can meet the latency, network size and power requirements for a wearable, wireless prosthesis/orthosis controller. Concomitant improvements in wireless communication and sensor technologies have increased capabilities of wearable biosensors. These improvements have not transferred to wireless prosthesis/orthosis controllers, in part due to strict latency and power consumption requirements. We used a Bluetooth Low Energy 5.3 (BLE) network to study the influence of the connection interval (10–100 ms) and event length (2500–7500 μs), ranges appropriate for real-time myoelectric prosthesis/orthosis control on the maximum network size, power consumption, and latency. The number of connections increased from 4 to 12 as the connection interval increased from 10 to 50 ms (event length of 2500 μs). For connection intervals ≤50 ms, the number of connections reduced by ≥50% with the increasing event length. At a connection interval of 100 ms, little change was observed in the number of connections vs. event length. Across event lengths, increasing the connection interval from 10 to 100 ms decreased the average power consumed by approximately 16%. Latency measurements showed that an average of one connection interval (maximum of just over two) elapses between the application of the signal at the peripheral node ADC input and its detection on the central node. Overall, reducing the latency using shorter connection intervals reduces the maximum number of connections and increases power consumption. [ABSTRACT FROM AUTHOR]

Deep neural networks (DNNs) and transfer learning (TL) have been used to improve surface electromyogram (sEMG)-based force estimation. However, prior studies focused mostly on applying TL within one joint, which limits dataset size and diversity. Herein, we investigated cross-joint TL between two upper-limb joints with four DNN architectures using sliding windows. We used two feedforward and two recurrent DNN models with feature engineering and feature learning, respectively. We found that the dependencies between sEMG and force are short-term (<400 ms) and that sliding windows are sufficient to capture them, suggesting that more complicated recurrent structures may not be necessary. Also, using DNN architectures reduced the required sliding window length. A model pre-trained on elbow data was fine-tuned on hand–wrist data, improving force estimation accuracy and reducing the required training data amount. A convolutional neural network with a 391 ms sliding window fine-tuned using 20 s of training data had an error of 6.03 ± 0.49% maximum voluntary torque, which is statistically lower than both our multilayer perceptron model with TL and a linear regression model using 40 s of training data. The success of TL between two distinct joints could help enrich the data available for future deep learning-related studies. [ABSTRACT FROM AUTHOR]

Recently, transformers have demonstrated notable improvements in natural advanced visual tasks. In the field of computer vision, transformer networks are beginning to supplant conventional convolutional neural networks (CNNs) due to their global receptive field and adaptability. Although transformers excel in capturing global features, they lag behind CNNs in handling fine local features, especially when dealing with underwater images containing complex and delicate structures. In order to tackle this challenge, we propose a refined transformer model by improving the feature blocks (dilated transformer block) to more accurately compute attention weights, enhancing the capture of both local and global features. Subsequently, a self-supervised method (a local and global blind-patch network) is embedded in the bottleneck layer, which can aggregate local and global information to enhance detail recovery and improve texture restoration quality. Additionally, we introduce a multi-scale convolutional block attention module (MSCBAM) to connect encoder and decoder features; this module enhances the feature representation of color channels, aiding in the restoration of color information in images. We plan to deploy this deep learning model onto the sensors of underwater robots for real-world underwater image-processing and ocean exploration tasks. Our model is named the refined transformer combined with convolutional block attention module (RT-CBAM). This study compares two traditional methods and six deep learning methods, and our approach achieved the best results in terms of detail processing and color restoration. [ABSTRACT FROM AUTHOR]

Navigation resource allocation for low-earth-orbit (LEO) constellations refers to the optimal allocation of navigational assets when the number and allocation of satellites in the LEO constellation have been determined. LEO constellations can not only transmit navigation enhancement signals but also enable space-based monitoring (SBM) for real-time assessment of GNSS signal quality. However, proximity in the frequencies of LEO navigation signals and SBM can lead to significant interference, necessitating isolated transmission and reception. This separation requires that SBM and navigation signal transmission be carried out by different satellites within the constellation, thus demanding a strategic allocation of satellite resources. Given the vast number of satellites and their rapid movement, the visibility among LEO, medium-earth-orbit (MEO), and geostationary orbit (GEO) satellites is highly dynamic, presenting substantial challenges in resource allocation due to the computational intensity involved. Therefore, this paper proposes an optimal allocation algorithm for LEO constellation navigation resources based on dynamic programming. In this algorithm, a network model for the allocation of navigation resources in LEO constellations is initially established. Under the constraints of visibility time windows and onboard transmission and reception isolation, the objective is set to minimize the number of LEO satellites used while achieving effective navigation signal transmission and SBM. The constraints of resource allocation and the mathematical expression of the optimization objective are derived. A dynamic programming approach is then employed to determine the optimal resource allocation scheme. Analytical results demonstrate that compared to Greedy and Divide-and-Conquer algorithms, this algorithm achieves the highest resource utilization rate and the lowest computational complexity, making it highly valuable for future resource allocation in LEO constellations. [ABSTRACT FROM AUTHOR]

Simple Summary: Ecdysone is an essential factor affecting host immunity, in addition to regulating insect metamorphosis. There are two isoforms of ecdysone receptors (EcR) in silkworm, BmEcR-B1, and BmEcR-A. In this study, BmEcR-B1 was found to be more critical than BmEcR-A in responding to BmNPV infection. Further studies confirmed that BmEcR-B1 played an anti-BmNPV role in the participation of 20E, which was associated with 20E-induced apoptosis. This study provides theoretical support for elucidating the resistant mechanism of silkworm to BmNPV and the control of BmNPV disease in sericulture. Bombyx mori nucleopolyhedrovirus (BmNPV) is a serious threat to sericulture. Nevertheless, no effective control strategy is currently available. The innate immunity of silkworm is critical in the antiviral process. Exploring its molecular mechanism provides theoretical support for the prevention and treatment of BmNPV. Insect hormone receptors play an essential role in regulating host immunity. We found a correlation between Bombyx mori ecdysone receptor B1 (BmEcR-B1) and BmNPV infection, whereas the underlying mechanism remains unclear. In this study, the expression patterns and sequence characteristics of BmEcR-B1 and its isoform, BmEcR-A, were initially analyzed. BmEcR-B1 was found to be more critical than BmEcR-A in silkworm development and responses to BmNPV. Moreover, RNAi and an overexpression in BmN cells showed BmEcR-B1 had antiviral effects in the presence of 20-hydroxyecdysone (20E); Otherwise, it had no antiviral activity. Furthermore, BmEcR-B1 was required for 20E-induced apoptosis, which significantly suppressed virus infection. Finally, feeding 20E had no significant negative impacts on larval growth and the cocoon shell, suggesting the regulation of this pathway has practical value in controlling BmNPV in sericulture. The findings of this study provide important theoretical support for understanding the mechanism of the silkworm innate immune system in response to BmNPV infection. [ABSTRACT FROM AUTHOR]

Wearable wireless biomedical sensors have emerged as a rapidly growing research field. For many biomedical signals, multiple sensors distributed about the body without local wired connections are required. However, designing multisite systems at low cost with low latency and high precision time synchronization of acquired data is an unsolved problem. Current solutions use custom wireless protocols or extra hardware for synchronization, forming custom systems with high power consumption that prohibit migration between commercial microcontrollers. We aimed to develop a better solution. We successfully developed a low-latency, Bluetooth low energy (BLE)-based data alignment method, implemented in the BLE application layer, making it transferable between manufacturer devices. The time synchronization method was tested on two commercial BLE platforms by inputting common sinusoidal input signals (over a range of frequencies) to evaluate time alignment performance between two independent peripheral nodes. Our best time synchronization and data alignment method achieved absolute time differences of 69 ± 71 μs for a Texas Instruments (TI) platform and 477 ± 490 μs for a Nordic platform. Their 95th percentile absolute errors were more comparable—under 1.8 ms for each. Our method is transferable between commercial microcontrollers and is sufficient for many biomedical applications. [ABSTRACT FROM AUTHOR]

Wireless wearable sensor systems for biomedical signal acquisition have developed rapidly in recent years. Multiple sensors are often deployed for monitoring common bioelectric signals, such as EEG (electroencephalogram), ECG (electrocardiogram), and EMG (electromyogram). Compared with ZigBee and low-power Wi-Fi, Bluetooth Low Energy (BLE) can be a more suitable wireless protocol for such systems. However, current time synchronization methods for BLE multi-channel systems, via either BLE beacon transmissions or additional hardware, cannot satisfy the requirements of high throughput with low latency, transferability between commercial devices, and low energy consumption. We developed a time synchronization and simple data alignment (SDA) algorithm, which was implemented in the BLE application layer without the need for additional hardware. We further developed a linear interpolation data alignment (LIDA) algorithm to improve upon SDA. We tested our algorithms using sinusoidal input signals at different frequencies (10 to 210 Hz in increments of 20 Hz—frequencies spanning much of the relevant range of EEG, ECG, and EMG signals) on Texas Instruments (TI) CC26XX family devices, with two peripheral nodes communicating with one central node. The analysis was performed offline. The lowest average (±standard deviation) absolute time alignment error between the two peripheral nodes achieved by the SDA algorithm was 384.3 ± 386.5 μs, while that of the LIDA algorithm was 189.9 ± 204.7 μs. For all sinusoidal frequencies tested, the performance of LIDA was always statistically better than that of SDA. These average alignment errors were quite low—well below one sample period for commonly acquired bioelectric signals. [ABSTRACT FROM AUTHOR]

Cross-modal vehicle localization is an important task for automated driving systems. This research proposes a novel approach based on LiDAR point clouds and OpenStreetMaps (OSM) via a constrained particle filter, which significantly improves the vehicle localization accuracy. The OSM modality provides not only a platform to generate simulated point cloud images, but also geometrical constraints (e.g., roads) to improve the particle filter's final result. The proposed approach is deterministic without any learning component or need for labelled data. Evaluated by using the KITTI dataset, it achieves accurate vehicle pose tracking with a position error of less than 3 m when considering the mean error across all the sequences. This method shows state-of-the-art accuracy when compared with the existing methods based on OSM or satellite maps. [ABSTRACT FROM AUTHOR]

LiDAR has been widely used in autonomous driving systems to provide high-precision 3D geometric information about the vehicle's surroundings for perception, localization, and path planning. LiDAR-based point cloud semantic segmentation is an important task with a critical real-time requirement. However, most of the existing convolutional neural network (CNN) models for 3D point cloud semantic segmentation are very complex and can hardly be processed at real-time on an embedded platform. In this study, a lightweight CNN structure was proposed for projection-based LiDAR point cloud semantic segmentation with only 1.9 M parameters that gave an 87% reduction comparing to the state-of-the-art networks. When evaluated on a GPU, the processing time was 38.5 ms per frame, and it achieved a 47.9% mIoU score on Semantic-KITTI dataset. In addition, the proposed CNN is targeted on an FPGA using an NVDLA architecture, which results in a 2.74x speedup over the GPU implementation with a 46 times improvement in terms of power efficiency. [ABSTRACT FROM AUTHOR]

To facilitate the broader use of EMG signal whitening, we studied four whitening procedures of various complexities, as well as the roles of sampling rate and noise correction. We separately analyzed force-varying and constant-force contractions from 64 subjects who completed constant-posture tasks about the elbow over a range of forces from 0% to 50% maximum voluntary contraction (MVC). From the constant-force tasks, we found that noise correction via the root difference of squares (RDS) method consistently reduced EMG recording noise, often by a factor of 5–10. All other primary results were from the force-varying contractions. Sampling at 4096 Hz provided small and statistically significant improvements over sampling at 2048 Hz (~3%), which, in turn, provided small improvements over sampling at 1024 Hz (~4%). In comparing equivalent processing variants at a sampling rate of 4096 Hz, whitening filters calibrated to the EMG spectrum of each subject generally performed best (4.74% MVC EMG-force error), followed by one universal whitening filter for all subjects (4.83% MVC error), followed by a high-pass filter whitening method (4.89% MVC error) and then a first difference whitening filter (4.91% MVC error)—but none of these statistically differed. Each did significantly improve from EMG-force error without whitening (5.55% MVC). The first difference is an excellent whitening option over this range of contraction forces since no calibration or algorithm decisions are required. [ABSTRACT FROM AUTHOR]

Mixed seeds of fused quartz particles and silicon particles were laid at the bottom of the crucible to assist the growth of multicrystalline silicon crystals. The full melting process was used, and then we found that the grown crystals had higher quality. The effect of mixed seeds on the growth of multicrystalline silicon was studied. The results showed that fine and uniform initial grains could be obtained by mixed seeds assisting the growth of crystals. Increasing the number of grain boundaries can better release thermal stress and inhibit the proliferation and diffusion of dislocations. The defect density of multicrystalline silicon decreased and the minority carrier lifetime increased, thus improving the conversion efficiency of multicrystalline silicon cells. [ABSTRACT FROM AUTHOR]

Blind spot detection is an important feature of Advanced Driver Assistance Systems (ADAS). In this paper, we provide a camera-based deep learning method that accurately detects other vehicles in the blind spot, replacing the traditional higher cost solution using radars. The recent breakthrough of deep learning algorithms shows extraordinary performance when applied to many computer vision tasks. Many new convolutional neural network (CNN) structures have been proposed and most of the networks are very deep in order to achieve the state-of-art performance when evaluated with benchmarks. However, blind spot detection, as a real-time embedded system application, requires high speed processing and low computational complexity. Hereby, we propose a novel method that transfers blind spot detection to an image classification task. Subsequently, a series of experiments are conducted to design an efficient neural network by comparing some of the latest deep learning models. Furthermore, we create a dataset with more than 10,000 labeled images using the blind spot view camera mounted on a test vehicle. Finally, we train the proposed deep learning model and evaluate its performance on the dataset. [ABSTRACT FROM AUTHOR]