Search

In this article, we propose an effective grasp detection network based on an improved deformable convolution and spatial feature center mechanism (DCSFC-Grasp) to precisely grasp unidentified objects. DCSFC-Grasp includes three key procedures as follows. First, improved deformable convolution is introduced to adaptively adjust receptive fields for multiscale feature information extraction. Then, an efficient spatial feature center (SFC) layer is explored to capture the global remote dependencies through a lightweight multilayer perceptron (MLP) architecture. Furthermore, a learnable feature center (LFC) mechanism is reported to gather local regional features and preserve the local corner region. Finally, a lightweight CARAFE operator is developed to upsample the features. Experimental results show that DCSFC-Grasp achieves a high accuracy (99.3% and 96.1% for the Cornell and Jacquard grasp datasets, respectively) and even outperforms the existing state-of-the-art grasp detection models. The results of real-world experiments on the six-DoF Realman RM65 robotic arm further demonstrate that our DCSFC-Grasp is effective and robust for the grasping of unknown targets. [ABSTRACT FROM AUTHOR]

Determining the quality of Ti-6Al-4V parts fabricated by selective laser melting (SLM) remains a challenge due to the high cost of SLM and the need for expertise in processes and materials. In order to understand the correspondence of the relative density of SLMed Ti-6Al-4V parts with process parameters, an optimized extreme gradient boosting (XGBoost) decision tree model was developed in the present paper using hyperparameter optimization with the GridsearchCV method. In particular, the effect of the size of the dataset for model training and testing on model prediction accuracy was examined. The results show that with the reduction in dataset size, the prediction accuracy of the proposed model decreases, but the overall accuracy can be maintained within a relatively high accuracy range, showing good agreement with the experimental results. Based on a small dataset, the prediction accuracy of the optimized XGBoost model was also compared with that of artificial neural network (ANN) and support vector regression (SVR) models, and it was found that the optimized XGBoost model has better evaluation indicators such as mean absolute error, root mean square error, and the coefficient of determination. In addition, the optimized XGBoost model can be easily extended to the prediction of mechanical properties of more metal materials manufactured by SLM processes. [ABSTRACT FROM AUTHOR]

The processing of natural wood into a high‐performance wood‐based composite to broaden the potential applications of wood can be achieved by various techniques using bulk wood as a raw material. However, attention has seldom been paid to the use of wood veneers for preparing high‐performance structural materials. In this study, delignified birch veneers are used as raw materials to fabricate high‐strength delignified wood‐based laminated composites (DWLCs) with phenol‐formaldehyde (PF) resin as adhesive. The DWLCs with a PF resin content of 40% had high mechanical properties and displayed flexural and tensile stress of 405.8 and 324.4 MPa, respectively. The flexural and tensile stress are 1.24 and 1.21 times higher than those of natural wood‐based laminated composites (WLCs). And we confirmed that partial delignification could contribute to enhancing the mechanical properties of high‐strength wood‐based composites. By using scanning electron microscopy (SEM), and energy‐dispersive spectroscopy (EDS), the strengthening mechanisms of the high‐performance DWLCs were investigated. Due to both the surface of the cell lumen and cell walls being filled with PF resin, the wood microstructure was cured under high pressure so that the DWLCs were enhanced. [ABSTRACT FROM AUTHOR]

High‐performance composites reinforced with natural fibers are next‐generation structural materials due to their recyclability and light weight. Manufacturing composites with high strength has been a challenge. Here, we describe the manufacture of novel composites reinforced with wood fibers using densification by hot‐pressing and fixing by curing reaction with phenolic resin. This composite with 1.4 g/cm3 had high mechanical properties, high water stability, and aging‐resistance, and exhibited a flexural strength and internal bonding strength of 102.4 and 7.5 MPa, respectively. The strength and modulus are 1.5–6.0 and 1.3–28.1 times higher than those of most of the wood fibers reinforced composites. After immersing in water for 24 h, the water absorption and thickness swelling are only 1.3% and 0.8%, respectively. Scanning electron microscope characterization indicates that the cell wall is totally collapsed and became complete densification. And the collapsed cells are fixed by phenolic resin. This production strategy for the composites was shown to be universally effective for various species of natural fibers. These novel wood fiber‐reinforced composites were suitable for commercial manufacture in the current production line, making them a low cost and high‐performance alternative to automotive and aerospace industries. [ABSTRACT FROM AUTHOR]

A Super-flexible and transparent wood film/silver nanowire electrode for optical and capacitive dual-mode sensing wood-based electronic skin. This work opens new directions for directly incorporating natural wood into e-skin sensors and may offer potential applications in wearable electronics, artificial intelligence, soft robots and so on. [Display omitted] • A wood-based e-skin with instrument- and naked eye-readable characteristics. • A pyramidal microstructure polydimethlysiloxane film was used as a sensing layer. • A super-flexible transparent wood film was used as the electrodes. • The e-skin can map the spatial distribution of static and dynamic pressures. • The e-skin in luminescence exhibited a maximum sensitivity value at 1.28 kPa−1. Flexible electronic skin (e-skin) with multimode broad-range pressure sensing and high sensitivity is highly desirable for robotics, artificial intelligence and human–machine interaction applications. Most research on e-skin has focused on improving the sensitivity of sensing layers interfaced with an electronic output device, with only a limited number of studies focused on user interfaces based on a human-readable output. Moreover, naturally degradable materials are receiving increasing attention for e-skin. Herein, inspired by a bioluminescent noctiluca, a wood-based e-skin with instrument- and naked eye-readable characteristics was sandwich assembled using silver nanowires and a super-flexible transparent wood (STW) film as an electrode and a pyramidal microstructure polydimethlysiloxane film as a stimuli-responsive layer. The electrode was prepared by depositing silver nanowires on the surface of the STW film through a Meyer rod coating method. The film demonstrated outstanding performance with high transmittance (91.4%), a small curvature radius (2 mm) and excellent sheet resistance stability even after greater than 1500 bending cycles. The e-skin showed notable changes in electroluminescence from 0 to 150 kPa (where humans feel pain) and the sensitivity remains larger than 0.15 kPa−1 over the broad pressure range. The sensitivity in capacitance exhibited high sensitivity at 1.01 kPa−1 in the low-pressure regime, meanwhile, the luminescence exhibited a maximum value at 1.28 kPa−1 at medium pressure. Furthermore, the e-skin accurately mapped the spatial distribution of static and dynamic pressures through optical sensing. This work not only improves the sustainable development of materials in electrodes but also illustrates the potential of utilizing natural wood in high value-added e-skin. [ABSTRACT FROM AUTHOR]

Highlights Natural fiber/resin composites exhibit poor fiber–matrix compatibility and easy combustion. In this study, jute‐fiber‐reinforced epoxy resin was used to prepare composite materials by adding marine shell powder, which is a natural non‐combustible inorganic filler that originates from a variety of sources. The morphology, composition, thermal stability, combustion performance, and mechanical properties of the prepared composites were tested using several characterization methods. The results show that the marine shell powder was evenly distributed in the epoxy resin and did not affect the adhesion between the epoxy resin and the jute fiber. When 20% marine shell powder (SMJF20) was added to the jute‐fiber‐reinforced epoxy resin composite, the pyrolytic residual content was 21.88% (60.2% higher than that without), and the maximum heat release rate was 350.75 kW/m2 (22.8% lower than that without). The total heat release and total smoke release of SMJF20 were also 13.9% and 30% lower than that of MJF, respectively. The addition of marine shell powder did not significantly affect the tensile strength, whereas the tensile modulus of SMJF20 increased by approximately 13.9%, the fracture strain decreased by approximately 13.4%, and the bending strength and modulus decreased by approximately 12.9% and 13.0%, respectively. Marine shell powder is an ideal biomass filler for jute‐fiber‐reinforced epoxy resins. Jute‐fiber‐epoxy composites fortified with non‐combustible marine shell powder. Effective filler distribution without compromising epoxy‐jute fiber adhesion. 20% marine shell powder significantly reduces heat release and smoke release. Tensile strength unaffected; tensile modulus and residual content increased. Marine shell powder proves ideal for enhancing epoxy resin with jute fibers. [ABSTRACT FROM AUTHOR]

Deep convolution neural networks have proven their powerful ability in comparing many tasks of computer vision due to their strong data learning capacity. In this paper, we propose a novel end-to-end denoising network, termed Fourier embedded U-shaped network (FEUSNet). By analyzing the amplitude spectrum and phase spectrum of Fourier coefficients, we find that low-frequency features of an image are in the former while noise features are in the latter. To make full use of this characteristic, Fourier features are learned and are concatenated as a prior module that is embedded into a U-shaped network to reduce noise while preserving multi-scale fine details. In the experiments, we first present ablation studies on the Fourier coefficients' learning networks and loss function. Then, we compare the proposed FEUSNet with the state-of-the-art denoising methods in quantization and qualification. The experimental results show that our FEUSNet performs well in noise suppression and preserves multi-scale enjoyable structures, even outperforming advanced denoising approaches. [ABSTRACT FROM AUTHOR]

Ovarian cancer is the seventh most common type of cancer and the eighth most common cause of cancer‑associated mortality among women. A number of studies have hypothesized that the expression status of certain genes may be used to predict prognosis in ovarian cancer. In the present study, the RNA expression data from next‑generation sequencing and the clinical information of 413 patients from The Cancer Genome Atlas dataset was downloaded to identify the association between gene‑expression level and the survival time of the patients with ovarian serous cystadenocarcinoma. A five‑gene model was predicted to be significantly associated with patient survival in ovarian serous cystadenocarcinoma by using random survival forests variable hunting algorithm and Cox analysis. A total of two genes, mesencephalic astrocyte‑derived neurotrophic factor and dedicator of cytokinesis 11, of the predicted five genes demonstrated positive expression in the ovarian serous cystadenocarcinoma cancer tissues by polymerase chain reaction analysis. Kaplan‑Meier and Receiver Operating Characteristic analysis confirmed that the model of the two genes exhibited high sensitivity and specificity to predict the prognostic survival of patients. In conclusion, the expression of the two genes in the two‑gene model was associated with the prognostic outcomes of patients with ovarian serous cystadenocarcinoma; the model demonstrated potential as a novel prognostic indicator, which may have important clinical significance. [ABSTRACT FROM AUTHOR]

Borophene, a new kind of two-dimensional materials, were successfully synthesized in experiment recently with potential applications. In this study, we have investigated the hydrogen storage performances of alkali-metal (Li, Na and K) doped three types of borophene polytypes synthesized on Ag substrate. It is found that strong binding strength exists between alkali-metal atoms and borophene, where metal atoms on borophene with separated distribution are energy more favorable than the formation of metal clusters, avoiding the aggregation problems. Polarization mechanism plays a dominant role in H 2 adsorption and the obtained storage capacities are closely related with the configurations of borophene, types of foreign atoms and the electronic interactions therein. The effects of temperature and pressure have also been taken into consideration through modified van’t Hoff equation. Our results demonstrate that Na-doped S2 and S3 and Li-doped three types of borophene could be severed as promising candidates for hydrogen storage. [ABSTRACT FROM AUTHOR]