Advanced intelligent systems (Weinheim an der Bergstrasse, Germany)最新文献

Deep learning (DL) has been very successful for classifying images, detecting targets, and segmenting regions in high-resolution images such as whole slide histopathology images. However, analysis of such high-resolution images requires very high DL complexity. Several AI optimization techniques have been recently proposed that aim at reducing the complexity of deep neural networks and hence expedite their execution and eventually allow the use of low-power, low-cost computing devices with limited computation and memory resources. These methods include parameter pruning and sharing, quantization, knowledge distillation, low-rank approximation, and resource efficient architectures. Rather than pruning network structures including filters, layers, and blocks of layers based on a manual selection of a significance metric such as l1-norm and l2-norm of the filter kernels, novel highly efficient AI-driven DL optimization algorithms using variations of the squeeze and excitation in order to prune filters and layers of deep models such as VGG-16 as well as eliminate filters and blocks of residual networks such as ResNet-56 are introduced. The proposed techniques achieve significantly higher reduction in the number of learning parameters, the number of floating point operations, and memory space as compared to the-state-of-the-art methods.

Most in vitro studies regarding new anticancer treatments are performed on 2D cultures, despite this approach imposes several limitations in recapitulating the real tumor behavior and in predicting the effects of therapy on both cancer and healthy tissues. Herein, advanced in vitro models based on scaffolds that support the 3D growth of glioma cells, further allowing the cocultures with healthy brain cells, are presented. These scaffolds, doped with superparamagnetic iron oxide nanoparticles and obtained through 2-photon polymerization, can be remotely manipulated thanks to an external magnet, thus obtaining biomimetic 3D organization recapitulating the brain cancer microenvironment. From a geometric point of view, the structure is functional to both cell culture on individual unit scaffolds and to tailored cocultures fostered by magnetic-driven unit assembly, also allowing for cell migration thanks to passages/fenestrations on adjacent structures. Leveraging magnetic dragging, for which a mathematical model is introduced, multiple cocultures are achieved, highlighting the high versatility and the user-friendly character of the proposed platform that can help overcome the current challenges in 3D cocultures handling, and open the way to the construction of increasingly biomimetic artificial systems.

Grey wolf optimizer (GWO) is a highly valued heuristic algorithm in many fields. However, for some complex problems, especially high-dimensional and multimodal problems, the basic algorithm has limited computational power and cannot get a satisfactory answer. In order to find a better solution, an improved algorithm based on GWO is proposed herein. Gaussian barebone, random selection and chaotic game mechanisms are introduced into the GWO algorithm to enhance the global search ability. The GWO enhanced by three mechanisms is called CBRGWO. To verify the performance of CBRGWO, using IEEE CEC 2017 as a test function, CBRGWO is compared to five GWO variants, five basic algorithms, six advanced algorithms, and four champion algorithms. CBRGWO is evaluated using the Friedman test and Wilcoxon signed-rank test. Then, the stability of CBRGWO is analyzed. To verify that CBRGWO is still effective in practical application, CBRGWO is applied to five engineering problems and a water quality prediction problem. The experimental findings indicate that CBRGWO maintains excellent optimization ability in practical engineering problems.

Small-scale magnetic robots are extensively recognized as promising untethered devices that can be controlled externally for numerous microscale applications. This study is proposed to address the independent control of multiple magnetic millirobots using an array of electromagnetic coils. Herein, each of the fabricated magnetic millirobots is magnetized with a dissimilar magnetization profile. Further, these millirobots are independently controlled using the mentioned magnetization strategy in addition to the supply of controlled current to each electromagnetic coil. To explore the physics of this combined stepwise approach in controlling the millirobots, theoretical and numerical investigations are carried out that further ensure the practical significance for broad applications. For demonstration purposes, three different shear-induced flow manipulation experiments, including the particle manipulation task, fluid color transition task, and micromixing task, are conducted using more than one millirobot with distinct motions. A maximum of three millirobots controlled with different motions are employed in the micromixing task, and further, it is observed to achieve nearly 80% mixing efficiency within 45 s. The presented work with the introduced actuation system and motion control strategies can strengthen the existing methods of small-scale robots for various applications, particularly for tasks that demand multiple millirobots.

The rapid advancement of soft robotic technology emphasizes the growing importance of tactile perception. Soft grippers, equipped with tactile sensing, can gather interactive information crucial for safe human–robot interaction, wearable devices, and dexterous manipulation. However, most soft grippers with tactile sensing abilities have limited modes of tactile perception, restricting their dexterity and safety. In addition, existing tactile systems are often complicated, leading to unstable perception signals. Inspired by various organisms, a novel multimodal tactile-sensing soft robotic finger is proposed. This finger, based on a modified fin ray structure, integrates a distributed fiber optic sensing system as part of its tactile sensory neural system. It replicates human finger capabilities, discerning contact forces as low as 0.01 N with exceptional sensitivity (106.96 mN nm⁻¹). Through training neural networks models, the finger achieves an accuracy exceeding 96% in recognizing roughness, material stiffness, and finger pad position. Assembled into two-finger parallel gripper, it demonstrates precise manipulation capabilities for fragile items like strawberries and potato chips. Moreover, through synergistic interplay of multimodal tactile sensing, this finger can successfully grasp an underwater transparent sphere, mitigating limitations of visual perception. The developed soft finger holds promise in various scenarios including hazardous environment detection and specialized grasping tasks.

Herein, a comprehensive review of the evolution of intelligent communication anti-jamming techniques is provided. First, a clear definition of the concept and elaboration on the inherent connotations and capability characteristics of intelligent communication anti-jamming is provided. Additionally, the initial construction of an intelligent communication anti-jamming system architecture is outlined. Subsequently, the development of intelligent communication anti-jamming is delved, tracing its progression from early-stage adaptive anti-jamming techniques to the more recent advancements in intelligent anti-jamming, which are primarily based on game theory and machine learning. Moreover, the latest research findings in this domain are thoroughly examined and the existing challenges and bottlenecks that hinder current research progress are highlighted. Finally, several viable research directions for future studies in the field of intelligent anti-jamming are proposed. The objective is to facilitate readers in gaining a comprehensive understanding of the concept, characteristics, and system architecture of intelligent communication anti-jamming. Additionally, it is aimed to provide a comprehensive overview of the developmental process and current state of research, thereby furnishing conceptual and theoretical support for the design of more effective and practical intelligent communication anti-jamming systems.

Soft pouch motors, engineered to mimic the natural movements of skeletal muscles, play a crucial role in advancing robotics and exoskeleton development. However, the fabrication techniques often involve multistage processes; they lack soft sensing capabilities and are sensitive to cutting and damage. This work introduces a new textile-based pouch motors with the capacity for biaxial actuation and capacitive sensory functions, achieved through the application of computerized knitting technology using ultrahigh molecular weight polyethylene yarn (Spectra) and conductive silver yarns. This method enables the rapid and scalable mass fabrication of robust pouch motors. The resulting pouch motors exhibit maximum lifting capacity of 10 kg, maximum contraction of 53.3% along the y-axis, and transverse extension of 41.18% along the x-axis at 50 kPa pressure. Finite element analysis closely matches the experimental data. The capacitance signals in relation to contraction motion are well suited for detecting air pressure levels and hold promise for applications requiring robotic control. Notably, it effectively elevates an ankle joint simulator at a 20° angle, highlighting its potential for applications such assisting individuals with foot drop. This study presents a practical demonstration of the soft ankle exosuit designed to provide lifting support for individuals facing this mobility challenge.

Flexible strain sensors based on textiles have attracted extensive attention owing to their light weight, flexibility, and comfort when wearing. However, challenges in integrating textile strain sensors into wearable sensing devices include the need for outstanding sensing performance, long-term monitoring stability, and fast, convenient integration processes to achieve comprehensive monitoring. The scalable fabrication technique presented here addresses these challenges by incorporating customizable graphene-based sensing networks into knitted structures, thus creating sensing sleeves for precise motion detection and differentiation. The performance and real-world application potential of the sensing sleeve are evaluated by its precision in angle estimation and complex joint motion recognition during intra- and intersubject studies. For intra-subject analysis, the sensing sleeve only exhibits a 2.34° angle error in five different knee activities among 20 participants, and the sensing sleeves show up to 94.1% and 96.1% accuracy in the gesture classification of knee and elbow, respectively. For inter-subject analysis, the sensing sleeve demonstrates a 4.21° angle error, and it shows up to 79.9% and 85.5% accuracy in the gesture classification of knee and elbow, respectively. An activity-guided user interface compatible with the sensing sleeves for human motion monitoring in home healthcare applications is presented to illustrate the potential applications.

Accurate prediction of the future evolution of observational time series is a paramount challenge in current data-driven research. While existing techniques struggle to learn useful representations from the temporal correlations, the high dimensionality in spatial domain is always considered as obstacle, leading to the curse of dimensionality and excessive resource consumption. This work designs a novel structure-aware reservoir computing aiming at enhancing the predictability of coupled time series, by incorporating their historical dynamics as well as structural information. Paralleled reservoir computers with redesigned mixing inputs based on spatial relationships are implemented to cope with the multiple time series, whose core idea originates from the principle of the celebrated Granger causality. Representative numerical simulations and comparisons demonstrate the superior performance of the approach over the traditional ones. This work provides valuable insights into deeply mining both temporal and spatial information to enhance the representation learning of data in various machine learning techniques.

Rapid deployment of automation in today's world has opened up exciting possibilities in the realm of design and fabrication of soft robotic grippers endowed with sensing capabilities. Herein, a novel design and rapid fabrication by 3D printing of a mechano-optic force sensor with a large dynamic range, sensitivity, and linear response, enabled by metamaterials-based structures, is presented. A simple approach for programming the metamaterial's behavior based on mathematical modeling of the sensor under dynamic loading is proposed. Machine learning models are utilized to predict the complete force–deformation profile, encompassing the linear range, the onset of nonlinear behavior, and the slope of profiles in both bending and compression-dominated regions. The design supports seamless integration of the sensor into soft grippers, enabling 3D printing of the soft gripper with an embedded sensor in a single step, thus overcoming the tedious and complex and multiple fabrication steps commonly applied in conventional processes. The sensor boasts a fine resolution of 0.015 N, a measurement range up to 16 N, linearity (adj. R²–0.991), and delivers consistent performance beyond 100 000 cycles. The sensitivity and range of the embedded mechano-optic force sensor can be easily programmed by both the metamaterial structure and the material's properties.