Developing robust Alzheimer's Disease (AD) classification models necessitates extensive training data, but aggregating multi-center medical data poses privacy risks. Although Federated Learning (FL) and Swarm Learning (SL) allow training generic models without data sharing, their performance is limited by variations in AD pathology features and sample class imbalances across centers. To address this issue, we propose a novel Harmonized Swarm Learning framework with Guided Optimization (HSGO) to enhance multi-center collaboration while preserving data privacy. Our framework employs a class-balanced loss function to train a robust generic model and guides the optimization of personalized models towards the generic model, eliminating extra AD pathology feature extraction steps. Furthermore, we design a dynamic feature similarity storage mechanism to facilitate personalized training. Experiments performed under two different multi-center data partitioning scenarios demonstrate that HSGO achieves competitive performance when compared with five baseline methods. Additionally, Layer-wise Relevance Propagation (LRP) analysis indicates that HSGO may help identify potential key brain regions in AD by integrating local and global features compared to traditional SL.

This paper proposes a configurable and scalable framework based on Controlled Robot Language with Frame Semantics (FrameCRL) for plan generation. Given natural language instructions, FrameCRL constructs an equivalent formal semantic formulation in the form of discourse representation structures (DRS). Imperative verbs are extracted from the semantic structures as keys to anchor relevant semantic frames from FrameNet, and the selected semantic frames are used to construct goal statements in planning language. Non-imperative statements are further analyzed to generate object specifications and the initial state of the planning problem. These generated statements are then merged into a single planning script, which can be solved directly by the integrated planner. The performance of FrameCRL was evaluated on various natural language corpora and compared with large language models (LLM) based methods in plan generation. The results demonstrated the outperformance of FrameCRL in generating high-quality plans and its capability to handle large context scenarios. The FrameCRL was also tested on pick-and-place tasks using a dual-arm robot and it showcased a robust performance in linguistic understanding.

Colorectal cancer is one of the malignant tumors with high morbidity and mortality rates worldwide. Colonoscopy is currently the most effective clinical method for screening colorectal polyps and is of great significance for the early detection and intervention of the disease. It is crucial to accurately segment the polyp area from colonoscopy images. However, the significant variation in polyp appearance and the blurred boundaries with surrounding mucosa make accurate and robust automatic segmentation highly challenging. To address this, we propose a feature fusion network guided by multi-scale convolutional attention, named MsCAFF (Multi-scale Convolutional Attention-based Feature Fusion). Specifically, the Convolutional Attention Multi-scale Fusion module (CAMF) fuses multi-scale information from high-level feature layers based on convolution operations. It enhances the spatial representation of the feature map through channel and spatial attention mechanisms from CBAM (Convolutional Block Attention Module), while generating a global feature map to serve as the initial guidance for the RA (Reverse Attention) module. This design allows the model to better focus on the target when roughly locating polyp regions, without losing critical boundary information. Through quantitative and qualitative evaluations on multiple challenging datasets, the results show that our network MsCAFF achieves significant improvement in segmentation accuracy.

Wireless networked Multi-Agent Systems have been playing an increasingly important role in a variety of industrial applications, such as robotics and smart grids. It mainly leverages wireless communication to coordinate actions and share information among multi-agents. However, the traditional wireless networked multi-agent system relies on accurate system models, which have limited applicability in practical industrial systems. To solve this issue, this paper proposes a novel EstimationControl-Scheduling Co-design method for wireless networked Multi-Agent Systems (ECS-CoMASs). In ECS-CoMASs, state estimation, control policy, and resource scheduling are jointly considered to optimize the system performance subject to limited radio resources. Besides, a Deep Reinforcement Learning (DRL) strategy that incorporates historical information is proposed to obtain a deep network model that could respond to varying environment. The proposed method outperforms existing methods in fully cooperative tasks in terms of task completion time and success rate.

To address the reliability requirements of aeroengine blades under extreme conditions of high temperature, high pressure, and severe aerodynamic loads, this study presents the PointCurNor network architecture based on edge awareness. It targets the technology bottleneck of limited measurement accuracy due to under-segmentation in blade surface roughness detection. The network integrates multi-scale curvature features and normal vector representations, creating a decoupling mechanism for sub-component boundaries. By using an equal block-based algorithm for abnormal region removal, the method effectively suppresses noise from areas with abrupt curvature changes, achieving micrometer level measurement accuracy. Experimental results show obvious improvements in segmentation accuracy at complex interfaces, ensuring precise roughness measurement and enhancing overall detection efficiency.

The early screening of cervical cancer is of great significance to reduce the mortality rate, but the traditional manual reading has the problems of strong subjectivity and low efficiency. In this paper, a dual-resolution collaborative classification method based on morphological and color features is proposed. The nucleus is accurately segmented by an adaptive active contour model, and a multi-level Adaboost-SVM classifier is designed by combining morphological features (area, ellipse fitting) at low magnification and color features (hue, optical density) at high magnification to achieve automatic screening of abnormal cells. The experimental results show that the sensitivity of the proposed method is 88.9% and the specificity is 100% in 54 clinical smears, which significantly improves the recognition accuracy of abnormal cells and provides an effective solution for intelligent screening of cervical cancer.

The early screening of cervical cancer relies on the accurate identification of cell nuclei in microscopic images, but the cells are distributed in different focal planes under high magnification, resulting in local blur in a single frame image. Traditional fusion algorithms are difficult to meet the real-time requirements due to high computational complexity, and deep learning-based methods are limited by hardware cost and data dependence. In this paper, according to the gray characteristics of Pap staining cervical nucleus (the pixel value in clear areas is lower), a lightweight multi-focus image fast fusion algorithm is proposed. The fusion image is directly synthesized by extracting the minimum gray value of the RGB channel of the multi-focus plane image pixel by pixel. In addition, a coarse-fine double-step search strategy is used to quickly locate the optimal focal plane, and a dynamic threshold mechanism is introduced to reduce redundant calculations. The experimental results demonstrate that the proposed algorithm achieves state-of-the-art multi-focus image fusion performance in cervical cytology applications.

Accurate detection of aero-engine blades is critical for aviation safety and maintenance. While industrial endoscopes enable efficient inspection of complex engine interiors, existing systems struggle with blade detection due to geometric variations, lighting changes, reflections, and occlusions. This paper proposes a novel spatial-temporal transformer model that enhances small blade detection via multi-scale feature extraction and improves edge robustness with an enhanced deformable DETR module, addressing defects like erosion or fractures. For video sequences, a unified spatial-temporal representation resolves counting errors (e.g., omissions/duplicates). Experiments on a custom inspection platform demonstrate superior accuracy and reliability over state-of-the-art methods.

Industrial Control Systems (ICSs) are the core of industrial production. Wireless technology, with its flexibility and adaptability, is catalyzing a transformative shift from traditional ICS to the advanced Industrial Wireless Control Systems (IWCSs). However, the openness of wireless media, high dynamics of the environment, and resource scarcity present unprecedented security challenges of high security defense costs and low detection inaccuracy for IWCS. State-of-the-art methods primarily treat ICS as a typical cyber-physical system, which focuses on security issues from the cyber and control domains, rather than the physical domain. As a result, they are unable to fully address the high dynamics of wireless channels and unknown attacks, ultimately failing to meet the stringent security requirements of industrial systems. To this end, this paper proposes a physical-domain whitelist as the final line of security defense leveraging the finite nature of the physical behavior space in industrial production systems. Moreover, a holistic cross-domain security-safety architecture is introduced, drawing inspiration from the integrated cyber-control-physical collaboration. In the proposed architecture, the top-down inherent security-safety defense and bottom-up risk backtracking form a close loop, which not only prevents unknown attacks but also facilitates rapid localization and response to attacks. In the experiment, the composite AGV scheduling control has been developed to verify the effectiveness of the architecture. Ultimately, the potential challenges of the cross-domain architecture for IWCS safety-security defense have been summarized.

In the construction industry, the integration of artificial intelligence (AI) and robotics has led to significant advancements in automating various tasks. One critical aspect is the intelligent navigation of automatic mobile robot (AMR) within construction sites, where dynamic environments pose challenges such as inaccurate robot state estimation, irregular and textureless obstacle detection. To solve these issues, this paper designs an AI enabled obstacle-aware AMR intelligent navigation approach while ensuring robot safety and efficiency. Specifically, the robot is equipped with the complementary RGB camera, inertial measurement unit (IMU) and wheel encoder to estimate the states (i.e., positions, orientations and velocities), which have been tightly fused in the optimization framework. Furthermore, the RGB-Depth images are jointly fed into the AI model. It combines of Mask-RCNN and multi-scale attention network, to detect and segment the obstacles, and estimate the corresponding relative depth. It should be noted that the system incorporates obstacle awareness mechanisms to dynamically switch the robot’s velocities in response to obstacles in the environment, ensuring smooth and collision-free movement. Experimental results demonstrate the effectiveness and robustness of the proposed AI-enabled navigation system in real-world construction scenarios, showcasing its potential to enhance productivity and safety in construction automation.

Fiber-reinforced composites (FRCs) are widely utilized across various sectors, due to their outstanding mechanical properties. The arrangement of fibers within these composites considerably influences their mechanical behavior. However, the state-of-the-art techniques on fiber orientation measurement are plagued by issues such as discontinuous boundaries and the imprecise measurement of finely oriented fibers. To this end, this work introduces a pioneering margin-aware cylindrical deep hough network (MAC-DHN) to solve these problems. The cylindrical Hough architecture, which acknowledges the$\pi$-periodicity of fiber orientations, is specifically crafted to address the problem of discontinuous boundaries. Furthermore, we design an innovative sample-wise reweighting strategy for the cross-entropy loss that enhances the differentiation between finely oriented fibers. This strategy lessens the loss related to samples with minimal prediction probability margins between the correct classification and the adjacent fine-grained categories. To comprehensively examine the fiber orientation measurement techniques in FRCs, a new dataset named FrCs orientation measurement dataset (FCOM) has been built, and the proposed method has been rigorously assessed on this dataset. Experimental results indicate that the proposed method surpasses existing techniques in terms of$F\!-\!\text{measure}$and mean absolute error (MAE), with respective scores of 0.981 and 0.219.

The pathogen of the ongoing coronavirus disease 2019 (COVID-19) pandemic is a newly discovered virus called severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Testing individuals for SARS-CoV-2 plays a critical role in containing COVID-19. For saving medical personnel and consumables, many countries are implementing group testing against SARS-CoV-2. However, existing group testing methods have the following limitations: (1) The group size is determined without theoretical analysis, and hence is usually not optimal. This adversely impacts the screening efficiency. (2) These methods neglect the fact that mixing samples together usually leads to substantial dilution of the SARS-CoV-2 virus, which seriously impacts the sensitivity of tests. In this paper, we aim to screen individuals infected with COVID-19 with as few tests as possible, under the premise that the sensitivity of tests is high enough. We propose an eXpectation Maximization based Adaptive Group Testing (XMAGT) method. The basic idea is to adaptively adjust its testing strategy between a group testing strategy and an individual testing strategy such that the expected number of samples identified by a single test is larger. During the screening process, the XMAGT method can estimate the ratio of positive samples. With this ratio, the XMAGT method can determine a group size under which the group testing strategy can achieve a maximal expected number of negative samples and the sensitivity of tests is higher than a user-specified threshold. Experimental results show that the XMAGT method outperforms existing methods in terms of both efficiency and sensitivity.

Industrial automation requires robot dexterity to automate many processes such as product assembling, packaging, and material handling. The existing robotic systems lack the capability to determining proper grasp strategies in the context of object affordances and task designations. In this paper, a framework of task-oriented dexterous grasping is proposed to learn grasp knowledge from human experience and to deploy the grasp strategies while adapting to grasp context. Grasp topology is defined and grasp strategies are learned from an established dataset for task-oriented dexterous manipulation. To adapt to various grasp context, a reinforcement-learning based grasping policy was implemented to deploy different task-oriented strategies. The performance of the system was evaluated in a simulated grasping environment by using an AR10 anthropomorphic hand installed in a Sawyer robotic arm. The proposed framework achieved a hit rate of 100% for grasp strategies and an overall top-3 match rate of 95.6%. The success rate of grasping was 85.6% during 2700 grasping experiments for manipulation tasks given in natural-language instructions.

An automatic and accurate state recognition of switches on electrical cabinet control panels plays an increasingly important role in the routine inspection of power equipment. This paper presents a novel method for real-time cabinet panel switch state recognition using hybrid visual features. Compared to traditional methods, the proposed approach can ensure the rectangular object regions from images captured at arbitrary angles by applying the perspective transformation model. Besides, the switch regions are segmented and the corresponding visual features are extracted on HSV space, instead of raw RGB space, which overcomes the illumination variability issues. The morphological operations and the inherent geometrical constraints, are employed to group the switch regions. Eventually, the switch recognition is implemented on high-dimensional vector space using feature similarity discriminants. The proposed method has been evaluated on the image dataset collected from power station cabinets. The experimental results verify the effectiveness of the method.

The light-weight thermal-RGB-inertial sensing units are now gaining increasing research attention, due to their heterogeneous and complementary properties. A robust and accurate registration between a thermal-RGB camera and an inertial sensor is a necessity for effective thermal-RGB-inertial fusion, which is an indispensable procedure for reliable tracking and mapping tasks. This paper presents an accurate calibration method to geometrically correlate the spatial relationships between an RGB camera, a thermal camera and an inertial measurement unit (IMU). The calibration proceeds within the unified calibration framework (thermal-to-RGB, RGB-to-IMU). The extrinsic parameters are estimated by jointly optimizing both the chessboard corner reprojection errors and acceleration and angular velocity error terms. Extensive evaluations have been performed on the collected thermal-RGB-inertial measurements. In this experiments study, the average RMS translation and Euler angle errors are less than 6 mm and 0.04 rad respectively under 20% artificial noise.

The increasing demand for health informatics has become a far-reaching trend in the ageing society. The utilization of wearable sensors enables monitoring senior people daily activities in free-living environments, conveniently and effectively. Among the primary health-care sensing categories, the wearable visual-inertial modality for human motion tracking gradually exerts promising potentials. In this paper, we present a novel wearable heading estimation strategy to track the movements of human limbs. It adaptively fuses inertial measurements with visual features following locality constraints. Body movements are classified into two types: general motion (which consists of both rotation and translation). or degenerate motion (which consists of only rotation). A specific number of feature correspondences between camera frames are adaptively chosen to satisfy both the feature descriptor similarity constraint and the locality constraint. The selected feature correspondences and inertial quaternions are employed to calculate the initial pose, followed by the coarse-to-fine procedure to iteratively remove visual outliers. Eventually, the ultimate heading is optimized using the correct feature matches. The proposed method has been thoroughly evaluated on the straight-line, rotatory and ambulatory movement scenarios. As the system is lightweight and requires small computational resources, it enables effective and unobtrusive human motion monitoring, especially for the senior citizens in the long-term rehabilitation.

Combining light-weight visual and inertial modalities for motion capturing has been popular in robotics researches. There exist scale ambiguity, inaccurate pose estimation with little or no baseline, incremental drifts over time in visual-inertial fusion. Thus, in this paper, we propose a robust motion capturing method based on the multi-joint kinematic chain using monocular visual-inertial sensors. Our method is able to recover monocular visual scale through the joint geometry constraint. Additionally, we take inertial pre-integration to assist visual outlier removal using Maximum A Posteriori method. Ultimately, the kinematic chain model is leveraged to constrain the associated multiple visual-inertial estimation drifts during long time tracking. In the experiments, we conduct multi-joint capturing on a robotic arm. The quality of motion reconstruction is evaluated by comparing the estimated results with the measurements from an optical motion tracking system OptiTrack.

This paper presents an orientation estimation methods using inertial cues (IMU) and visual feature constraint. Our proposed approach combines both of these two modalities in an original way. Two feature-point correspondences between consecutive frames are firstly selected that not merely meet the requirement of descriptor similarity constraint but the locality constraint. Secondly, these two selected correspondences together with inertial quaternions are jointly employed to derive the initial body pose. Thirdly, a coarse-to-fine procedure proceeds in removing visual false matches and in estimating body poses iteratively using the Posteriori Bayes Rule and Expectation Maximization. Eventually, the optimal orientation is estimated via the iteratively selected visual inliers. Experimental results validate that our proposed strategy is effective and accurate in orientation estimate.