98%
921
2 minutes
20
Lossless and efficient robotic grasping is becoming increasingly important with the widespread application of intelligent robotics in warehouse transportation, human healthcare, and domestic services. However, current sensors for feedback of grasping behavior are greatly restricted by high manufacturing cost, large volume and mass, complex circuit, and signal crosstalk. To solve these problems, here, we prepare lightweight distance sensor-based reduced graphene oxide (rGO)/MXene-rGO coaxial microfibers with interface buffer to assist lossless grasping of a robotic manipulator. The as-fabricated distance microsensor exhibits a high sensitivity of 91.2 m in the distance range of 50-300 μm, a fast response time of 116 ms, a high resolution of 5 μm, and good stability in 500 cycles. Furthermore, the high-performance and lightweight microsensor is installed on the robotic manipulator to reflect the grasp state by the displacement imposed on the sensor. By establishing the correlation between the microsensing signal and the grasp state, the safe, non-destructive, and effective grasp and release of the target can be achieved. The lightweight and high-powered distance sensor displays great application prospects in intelligent fetching, medical surgery, multi-spindle automatic machines, and cultural relics excavation.
Download full-text PDF |
Source |
---|---|
http://dx.doi.org/10.1021/acs.langmuir.3c00374 | DOI Listing |
Sci Rep
September 2025
Fukushima Renewable Energy Institute, Koriyama, 963-0298, Japan.
This study proposes a novel and computationally efficient method for real-time identification and localization of power quality (PQ) disturbances in microgrids using dynamic Lissajous patterns formed by voltage and current waveforms. Each power disturbance-such as sag, swell, harmonic distortion, and transients-induces a unique geometric deformation in the Lissajous figure, which serves as a visual signature of the event. Key geometric and statistical features, including area, skewness, kurtosis, and centroid deviation, are extracted from these dynamic patterns to construct robust indices for classification.
View Article and Find Full Text PDFNeural Netw
August 2025
Inspur Electronic Information Industry Co., Ltd, China.
Knowledge distillation (KD) makes it possible to deploy high-accuracy models on devices with limited resources and is an effective means of achieving lightweight models. With the advancement of technology, the methods of knowledge distillation are also continuously developing and improving to adapt to different application scenarios and needs. To facilitate the transfer of knowledge from larger networks to smaller and lighter networks, KD has been employed to bridge the gap in probability outputs or middle-layer representations between teacher and student networks.
View Article and Find Full Text PDFComput Biol Med
September 2025
London South Bank University, Department of Computer Science & Informatics, 103 Borough Rd, London, SE1 0AA, United Kingdom.
Delineating and classifying individual cells in microscopy tissue images is inherently challenging yet remains essential for advances in medical and neuroscientific research. In this work, we propose a new deep learning framework, CISCA, for automatic cell instance segmentation and classification in histological slices. At the core of CISCA is a network architecture featuring a lightweight U-Net with three heads in the decoder.
View Article and Find Full Text PDFSci Rep
September 2025
College of Energy Engineering, Zhejiang University, Yuquan Campus, No. 38, Hangzhou, 310027, P.R. China.
Nonmetallic pipelines are promising for medium-short distance hydrogen transport due to their lightweight, corrosion resistance, and durability. However, their low conductivity raises electrostatic safety concerns, given hydrogen's exceptionally low ignition energy (0.017 mJ).
View Article and Find Full Text PDFJ Cheminform
August 2025
Bioinformatics Group, Wageningen University & Research, Droevendaalsesteeg 4, 6708 PB, Wageningen, the Netherlands.
Natural products provide a rich source of bioactive molecules for a variety of applications. Molecular fingerprints are the tool of choice for systematic large-scale studies of their structures. However, current molecular fingerprints insufficiently represent characteristic features of natural products inherently, decreasing the interpretability of natural product-specific predictions.
View Article and Find Full Text PDF