Band selection, aiming at screening representative spectral bands and eliminating redundant information, has long been a popular topic in hyperspectral imagery (HSI) processing, and has garnered a growing concern owing to the advancements in sparse representation techniques. Traditional sparsity-based methods are frequently impeded by the issue of inadequate or even unattainable training samples. Moreover, these approaches may fall short in thoroughly investigating the spatial structural information and spectral contextual information. To this end, this paper proposes a new unsupervised band selection scheme, namely, pseudo-label guided sparse regression with spatial and spectral regularization (PSR2BS), which embeds band selection into an unsupervised sparse regression model. Specifically, a pseudo-label matrix is jointly learned to serve as a discriminative cluster indicator, during which it guides the projection matrix in selecting informative bands. To leverage the spatial information within HSI, an image is segmented into different distinct homogeneous regions to generate representation samples, wherein the local structural information is also explored through spatial regularization. Furthermore, a spectral regularization term is introducing, making full use of prior information regarding similarity within spectral bands. To solve the proposed model, an effective and efficient iterative optimization algorithm is developed. Our extensive experiments on classification and anomaly detection across six real HSI datasets clearly demonstrate the superior performance of the proposed PSR2BS compared with state-of-the-art competitors.
An intelligent memory-based event-triggered impulsive control (METIC) scheme is proposed to address the stabilization problem for a class of nonlinear systems while accounting for exponential convergence, dynamic performance, and control frequency. The contribution of the scheme is the incorporation of weighted historical data into the triggering condition, using both fixed thresholds and adaptive thresholds based on Q-learning. By utilizing the system states at the two most recent triggering instants to construct the triggering condition, several exponential stability criteria are first established via an iterative approach. Then, in the general case in which additional historical states are included, a comparison system approach is employed to derive new stability conditions. Furthermore, to adaptively tune the event-triggering thresholds and improve system performance, a Q-learning-based optimization algorithm is developed, and a set of easily verifiable stability conditions is derived within the framework of switched system theory. For both fixed-threshold and adaptive-threshold cases, Zeno behavior is rigorously excluded through theoretical analysis. Finally, comparative simulation results are presented to demonstrate the effectiveness of the proposed method.
With the rapid advancement of artificial intelligence, multi-agent systems (MASs) are evolving from classical paradigms toward architectures built upon large foundation models (LFMs). This survey provides a systematic review and comparative analysis of classical MASs (CMASs) and LFM-based MASs (LMASs). First, within a closed-loop coordination framework, CMASs are reviewed across four fundamental dimensions: perception, communication, decision-making, and control. Beyond this framework, LMASs integrate LFMs to lift collaboration from low-level state exchanges to semantic-level reasoning, enabling more flexible coordination and improved adaptability across diverse scenarios. Then, a comparative analysis is conducted to contrast CMASs and LMASs across architecture, operating mechanism, adaptability, and application. Finally, future perspectives on MASs are presented, summarizing open challenges and potential research opportunities.
Read moreSquare piles play a crucial role as load-bearing components in transmission tower foundations, impacting the safety of power line systems under complex loading conditions due to their superior lateral resistance performance. However, existing research has predominantly focused on circular piles, with limited systematic investigation into the lateral bearing mechanisms of square piles. This study utilizes finite element analysis to develop a comprehensive full-scale model incorporating pile-soil interactions and soil spatial effects. The study systematically analyzes the effects of pile side length and embedment ratio on load-displacement curves, horizontal ultimate stage, and overturning rotation center. The results indicate that (1) Base on load-displacement curve evolution, plastic strain distribution and experimental specifications, under horizontal ultimate state in sandy soil, the critical displacement threshold at the pile head of square piles is 10 mm. (2) An increase in ultimate lateral bearing capacity by a factor of 2.4 with the square pile side length increasing from 1.2 m to 2.0 m, and an increase in bearing capacity by 0.9 times with the embedment ratio rising from 1.0 to 2.0. (3) The overturning rotation center is located at the bottom central axis position of the square pile. (4) Under horizontal ultimate loading conditions, square piles induce lateral earth pressure in sandy soil that follows a parabolic distribution along the embedding depth, with peak stress occurring at 0.5 times the pile's embedding depth below the soil surface. The results provide theoretical and practical references for optimizing the seismic and disaster-resistant design of transmission infrastructure.
Read moreTactile sensing is a fundamental modality for embodied intelligence, offering unique and direct feedback on contact geometry, material properties, and interaction dynamics that remote sensors cannot replace. However, unimodal tactile perception is inherently limited by its sparse spatial coverage and lack of global semantic context. With the recent explosion in deep learning and large language models, integrating tactile with vision and language has become essential to bridge physical interaction with semantic reasoning, leading to the emergence of Multimodal Tactile Fusion. Despite rapid progress, the existing researches remain fragmented across disparate datasets, sensing modalities, and tasks, lacking a unified theoretical framework. To address this gap, this paper provides a comprehensive survey of multimodal tactile fusion research up to the first quarter of 2026. We propose a hierarchical taxonomy that organizes the field into two primary dimensions: multimodal datasets and multimodal methods. On the data side, we categorize resources ranging from Tactile-Vision datasets, Tactile-Language datasets, Tactile-Vision-Language datasets, and Tactile-Vision-Other datasets. On the method side, we structure prior work into three core pillars: (1) Multimodal Perception and Recognition, which focuses on object understanding and grasp prediction; (2) Cross-Modal Generation, focusing on bidirectional translation between tactile, vision, and text; and (3) Multimodal Interaction, emphasizing feedback control and language-guided manipulation. Furthermore, we summarize representative tactile sensing hardware, review commonly used evaluation metrics and benchmark settings, and discuss current challenges and promising future directions.
Read moreReferring Multi-Object Tracking (RMOT) aims to achieve precise object detection and tracking through natural language instructions, representing a fundamental capability for intelligent robotic systems. However, current RMOT research remains mostly confined to ground-level scenarios, which constrains their ability to capture broad-scale scene contexts and perform comprehensive tracking and path planning. In contrast, Unmanned Aerial Vehicles (UAVs) leverage their expansive aerial perspectives and superior maneuverability to enable wide-area surveillance. Moreover, UAVs have emerged as critical platforms for Embodied Intelligence, which has given rise to an unprecedented demand for intelligent aerial systems capable of natural language interaction. To this end, we introduce AerialMind, the first large-scale RMOT benchmark in UAV scenarios, which aims to bridge this research gap. To facilitate its construction, we develop an innovative semi-automated collaborative agent-based labeling assistant (COALA) framework that significantly reduces labor costs while maintaining annotation quality. Furthermore, we propose HawkEyeTrack (HETrack), a novel method that collaboratively enhances vision-language representation learning and improves the perception of UAV scenarios. Comprehensive experiments validated the challenging nature of our dataset and the effectiveness of our method.
Read moreThis study investigated the effects of low and high polarity extracts (LPE and HPE) of Angelica sinensis (AS) on growth, body composition, and metabolism in carp ( Cyprinus carpio var. Jian). Over a 42‐day period, 780 fish were randomly separated into 13 groups with three replicate aquariums respectively. Thirteen groups received the feeding of basic diet, six LPE diets, and six HPE diets, respectively. The results indicated that dietary LPE decreased weight gain (WG), condition factor (CF), the activities of Na + , K + ‐ATPase, γ‐glutamyl transpeptidase (γ‐GT), trypsin and lipase in digestive organs, plasma total amino acids (TAAs), triglyceride (TG) and ammonia levels, lipid productive value (LPV), and ammonia excretion rate (AER; p < 0.05), increased the activities of alkaline phosphatase (AKP), glutamate‐oxaloacetate transaminase (GOT), glutamate‐pyruvate transaminase (GPT), and α‐amylase in digestive organs, the content of total protein (TP) and glucose (GLU) in plasma, protein productive value (PPV), oxygen consumption rate (OCR), and O:N ratio in carp ( p < 0.05). At the same time, dietary HPE increased WG, CF, the activities of lactate dehydrogenase (LDH), trypsin and lipase in digestive organs, the content of TP in plasma and LPV, PPV, OCR, and O:N ratio ( p < 0.05) and decreased the activities of GOT and GPT in hepatopancreas as well as the content of ammonia, TAA, and GLU in plasma of carp ( p < 0.05). According to the above findings, dietary LPE inhibits the growth and accumulation of body lipid and enhances the accumulation of body proteins by decreasing the digestion and absorption of lipids as well as amino acid catabolism, and increasing the catabolism of sugar and fat in fish. Dietary HPE enhances the growth and accumulation of body lipid and proteins by decreasing the catabolism of amino acid and increasing the digestion and absorption of proteins and lipids and the catabolism of sugar in fish.
Read moreAfter laser powder bed fusion (LPBF) of an ultra-strong in situ TiC whisker reinforced β-Ti composite, this paper investigates the evolution of microstructure and mechanical properties in response to heat treatment at different temperatures. Using in depth nano-SEM and TEM analyses, it is shown that ageing at 400 °C rounds the whiskers, annihilates the strain fields and grows Mo segregated nano-cells, but without improving the ductility. In contrast, ageing at 600 °C enables the transformation of metastable β to a lamellar β + α, leading to a dual phase matrix embedding TiC particles. This is in such a manner that extra ageing at 600 °C coalesces the nano-lamellar α + β microstructure to form a coarser micro-lamellar α + β matrix. This microstructure achieves 66 % of the compressive deformation of Cp-Ti, and over 1400 MPa compressive strength after 1 h of ageing at 600 °C. Despite this success under compression, hard and stiff TiC particles may still cause large spherical fractured voids, severely limiting the plastic deformation under tension.
Read more
.jpg){kind=small}
