Respiratory rate is an important vital sign that can be used to determine human physiological state. In recent years, Wi-Fi-based contactless respiration monitoring has drawn significant attention due to the prevalence of wireless local area network (WLAN) infrastructure. Most existing approaches to respiration monitoring perform well in controlled environments, without the presence of additional moving individuals in the area of interest. A few recent studies have attempted to reduce the impact of other people moving in the vicinity of the target individual. However, these approaches exhibit notable limitations, such as restricting the number of interfering individuals to one, or requiring a direct wired connection between the Wi-Fi transmitter and receiver for synchronization. To address these issues, in this study, we develop a contactless respiration monitoring system using commodity Wi-Fi devices, which we name RoSense. Through a series of empirical studies, we observe that the channel state information (CSI) for subcarriers is significantly affected by the presence of interfering individuals, but a small subset retain relatively clear signal patterns linked to the target’s respiration. Leveraging these findings, RoSense employs a signal power-based subcarrier selection strategy to identify high-quality subcarriers. The selected subcarriers are then aligned to enhance signal gain and fused to complement the weaker periodic parts. Additionally, RoSense periodically detects the quality of subcarriers, selecting the most effective subcarriers to maximize the contribution of high-quality ones. Extensive experiments were performed in real-world settings with 10 volunteers to verify the feasibility and effectiveness of RoSense. Our results demonstrate that RoSense is able to achieve robust respiration monitoring by suppressing the impact of interfering individuals.

This paper proposes a resilient cooperative control framework for unmanned surface vehicles (USVs) operating under unknown disturbances, input saturation, and communication link faults. A novel resilient distributed observer is developed to simultaneously reconstruct both the state and system matrix of the leader using only local neighbor information without requiring global knowledge of the leader’s dynamics or communication topology as in existing methods. Furthermore, a disturbance-rejection synchronization controller is designed based on the internal model principle and a saturation-compensating auxiliary system. In contrast to conventional strategies that rely on constant-bound disturbance assumptions and lack explicit saturation compensation, the proposed controller achieves online disturbance cancellation without prior knowledge of disturbance characteristics such as frequency, amplitude, and phase, while explicitly handling input saturation. The proposed control strategy could significantly improve tracking accuracy, robustness, and practicality in marine environments. Theoretical analysis and simulations validate the merits and performance of the proposed approach.

This paper presents a novel integrated estimation-control framework to solve the path following problem for autonomous ground vehicles under coupled uncertainties in mass and cornering stiffness. The proposed method is characterized by the synergistic integration of three key innovations: 1) a fuzzy confidence-enhanced recursive least-squares algorithm (FC-RLS-MFF) that employs a rule-based mechanism to dynamically assess excitation sufficiency, ensuring robust online mass estimation; 2) a tightly-coupled identification scheme, where the estimated mass actively refines cornering stiffness in real time through tire vertical load updates, departing from the use of fixed nominal values; and 3) an adaptive fast nonsingular terminal sliding mode controller (AFNTSMC) that leverages this real-time parameter stream to enable model-based compensation, rather than mere disturbance rejection. Co-simulations, including ablation studies, validate the framework and demonstrate that this tight integration yields emergent performance gains, such as a significant reduction in path-tracking error.

Given the pervasive presence of feedback mechanisms and inertial loops inherent in industrial processes, increasing research efforts have focused on integrating latent feature dynamics into deep learning architectures to address the issue of strong dynamic autocorrelation in industrial processes. However, the complex temporal dependencies present in latent features and the limitations imposed by traditional alternating training methods have already constrained the performance of such models. In this paper, we propose a latent probabilistic dynamics embedding supervised deep networks for soft sensing. In detail, a probabilistic dynamic model with graph-guiding based on the past and current latent features is constructed in the latent space of the supervised deep networks to capture the complex dependencies between the latent sequences. The article introduces a new approach involving a probability-distribution-based predictive regularization term for latent features. By jointly training the model, the network parameters are optimized to ensure the overall convergence of the network. An improved variational graph recurrent neural network with the inclusion of randomness into the high-level latent space was proposed to model the latent dynamics for supervised deep networks and additional graph structure information to help analyze temporal dependencies. Finally, the proposed methods are implemented on two real industrial cases to demonstrate their effectiveness and superiority. Comparative experiments and ablation experiments are designed to illustrate the higher prediction accuracy and effectiveness of the proposed method.

Segment Anything Model (SAM) is a popular vision foundation model that can segment data from any domain. Benefiting from its outstanding generalization ability, SAM has been widely adopted in many industrial scenarios. However, as SAM is built upon a heavy Vision Transformer (ViT), it suffers from memory-hungry and low latency, which restricts the deployment on edge devices. In this paper, we systematically explore how to compress SAM effectively, making it feasible to adapt edge devices with limited calculation abilities. Specifically, our method consists of three aspects: weight initialization, knowledge distillation, and model quantization. It is notable that all three aspects are not simply inherited from previous methods, but tactfully designed based on the teacher-student learning paradigm, considering the task-attributes of SAM pre-training. Firstly, we design a weight initialization method for fully using the pre-training knowledge implicitly contained in the teacher’s parameter space. Secondly, based on the weight initialization, we design a novel distillation method tailored to SAM pre-training, focusing on learning the semantic differences among areas. Lastly, we perform quantization on our distilled models. Unlike the previous method, we used both the teacher and the student to calibrate our model in the quantization process. We conduct systematic experiments on various teacher-student network pairs to validate the broad effectiveness of our method. By applying our method, our target models achieve over 64.5× speed increase compared to the original SAM. Core code is available at: https://github.com/ZG-ZZ/DC-SAM.

ABSTRACT Metric learning is fundamental to multi‐view geo‐localization, as it aims to establish a distance metric that minimizes the feature space distance between similar data points while maximizing the separation between dissimilar ones. However, in Siamese networks employed for metric learning, individual branches may exhibit discrepancies in their interpretation of semantic information from input data, resulting in semantically inconsistent feature representations. To address this issue, a method is designed to enhance significant region consistency within multi‐view spaces by integrating feature consistency enhancement (FCE) and multi‐level feature mining (MLFM) techniques into a dual‐branch network. The FCE method emphasizes critical components of the input data, ensuring feature consistency between the two branches. Additionally, the MLFM mechanism facilitates feature integration across multiple levels, thereby enabling a more comprehensive extraction of semantic information. This approach enhances semantic understanding and promotes feature consistency across branches. The proposed method achieves AP values of 82.38% for drone‐to‐satellite and 77.36% for satellite‐to‐drone image matching. Notably, the method maintains computational efficiency without significantly affecting inference time. Additionally, improvements are observed in R@1, R@5 and R@10 metrics. The experimental results show that integrating FCE and MLFM into the dual‐branch network improves semantic representation and outperforms existing methods.

Multiplayer game theory has been widely studied, with most existing research focusing on fully connected network structures. In contrast, multiplayer graphical games consider sparser communication topologies, making them more practical for large-scale systems. This article, based on a reinforcement learning (RL) method, investigates the problem of computing Nash equilibrium (NE) strategies in a class of multiplayer graphical games where the system is influenced by an external system. To estimate the unknown states of the external system, we propose a distributed adaptive observer and prove that its observation error asymptotically converges to zero. Furthermore, we derive a range of discount factor values that preserve system stability. To solve for the NE strategy, we develop an off-policy algorithm integrated with the distributed adaptive observer for policy evaluation. To enhance convergence speed, we introduce a distributed policy improvement mechanism, which ensures policy convergence to equilibrium while maintaining system stability. The effectiveness of the proposed algorithm is validated through simulations on a voltage synchronization system.

The article studies the fully distributed leader–follower and adaptive event-triggered problem with the guarantee of positive minimum interevent times (MIET) for time-varying MAS on directed graphs. First, a novel fully distributed adaptive event-triggered scheme that includes a time-varying matrix gain and two dynamic gains is designed, and the requirement for global topology information can be removed. A novel dynamic triggering mechanism is then put forward for each follower, where an auxiliary dynamic parameter is introduced into the triggering function to guarantee the existence of positive MIETs. Meanwhile, it is worth mentioning that a direct measurement parameter using only the sampling information is leveraged to avoid the use of continuous communication with neighbors. Finally, simulation results are presented to confirm the performance of the proposed method.

This article proposes an event-based asymptotic tracking control method for nonlinear strict-feedback systems with multiple unknown control directions. The system is characterized by multiple unknown control directions, which pose challenges to its performance. In contrast to traditional Nussbaum-type methods, we propose a novel Nussbaum-type function to handle multiple Nussbaum-type gains, ensuring robust asymptotic tracking. Additionally, two event-triggered mechanisms are developed to alleviate the computational complexity of adaptive Nussbaum design. The static event-triggered mechanism significantly improves the system’s responsiveness to dynamic changes by employing dynamically decreasing thresholds. Building on this, a dynamic event-triggered mechanism is introduced, incorporating an internal variable that continuously adjusts the triggering conditions over time. Furthermore, the proposed design not only achieves asymptotic tracking control but also ensures that both event-triggered mechanisms avoid the Zeno phenomenon. To validate the proposed design schemes, a simulation example of a marine surface vehicle is presented.

The extraction of spatiotemporal neuron activity from calcium imaging videos plays a crucial role in unraveling the coding properties of neurons. While existing neuron extraction approaches have shown promising results, disturbing and scattering background and unused depth still impede their performance. To address these limitations, we develop an automatic and accurate neuron extraction paradigm, dubbed as decomposition–estimation–reconstruction (DER), consisting of D-procedure, E-procedure, and R-procedure. Specifically, the D-procedure first decomposes the raw data into a low-rank background and a sparse neuron signal, and regularizes$L_{0}$-norm priors of intensity and gradient of the neuron signal to suppress blurring and artifact effects. Then, the E-procedure estimates the depth-dependent transmission of the neuron signal based on its bright and dark channel priors. The R-procedure finally integrates the depth estimation of the neuron signal as a content-importance weight into a constrained non-negative matrix decomposition framework, which facilitates accurate neuron locations to boost the quality of extracted neurons. These three procedures are coupled in a cascade manner, where the former copes with calcium imaging data to facilitate the subsequent one. Comprehensive experiments on neuron extraction from calcium imaging videos demonstrate the superiority of our DER paradigm in both qualitative results and quantitative assessments over state-of-the-art methods.

In recent years, WiFi-based device-free localization (DFL) has attracted attentions due to the rapid development of location-based applications. The localization performance of data-driven DFL models highly relies on the quality of the fingerprint. However, it is difficult to obtain high-quality fingerprints in cluttered environments due to various uncertainties, such as environmental dynamics. To mitigate effects of uncertainties, this article proposes a variance-constrained local–global modeling method to enhance the localization performance. To be specific, the collected channel state information data from a specific environment are first divided into several groups depending on their statistical characteristics using the clustering method, and the local–global modeling mechanism is then implemented based on the extreme learning machine, a kind of noniterative single hidden layer feedforward neural network, to achieve the good representation of the whole environment. During the local–global modeling process, the proposed method is to simultaneously minimize the output weights of the neural network, training errors of the DFL model, and intragroup variances of the grouped data, which makes the created DFL model robust in cluttered environments and not sensitive to uncertainties. Comprehensive experiments in several scenarios, including different indoor environments, device positions and heights, target's orientations, and body shapes, are performed. Experimental results indicate that the proposed method could achieve better localization performance than selected baseline methods, demonstrating the effectiveness of the proposed variance-constrained local–global modeling mechanism in DFL.

Human Identification at a Distance (HID) is an important research area due to its importance (especially in biometrics) and inherent challenges within this domain. To mitigate some of the constraints, we have introduced the HID challenge. This paper presents an overview of the 4th International Competition on Human Identification at a Distance (HID 2023), which serves as a benchmark for evaluating various methods in the field of human identification at a distance. We have introduced a new dataset, SUSTech-Competition, engulfing a cross-domain challenge. This dataset has 859 subjects, having various variations of clothing, carrying conditions, occlusions, and view angles. With a substantial participation of 254 registered teams, HID 2023 has attracted considerable attention and yielded highly encouraging results. Notably, the top-performing teams achieved significantly good accuracies. In this paper, we provide an introduction to the competition, encompassing the dataset, experimental settings, and competition organization, as well as an analysis of the results obtained by the top teams. Additionally, we delve into the methodologies employed by these leading teams. The progress demonstrated in this competition offers an optimistic outlook on the advancements in gait recognition, highlighting its potential for robust real applications.

This article considers the design of an adaptive iterative learning controller for high-rise buildings with active mass dampers (AMDs). High-rise buildings in this article are seen as distributed parameter systems, in which the characteristics of every point in buildings should be considered. Two partial differential equations (PDEs) and several ordinary differential equations are used to describe the model of buildings. To achieve the control target that is to suppress the vibration induced by high winds, an adaptive iterative learning controller is proposed for the flexible building system with boundary disturbance. The convergency of the adaptive iterative learning control (AILC) approach is proven by serious theory analysis. In simulations and experiments, this article uses both the analysis of figures and quantitative analysis (root-mean-square values) to illustrate the efficiency of the AILC scheme.

The fully distributed adaptive event-triggered average tracking protocol for general linear multiagent systems (MASs) with a positive minimum interevent time (PMIET) is established in an asynchronous way. First, a fully distributed event-triggered detector and controller for each agent that includes internal variables and adaptive gains are designed to track the average of multiple time-varying references signals asymptotically via intermittent communication without the requirement of global information. Then, activable dynamic event-triggering mechanisms are designed for the detector and the controller, respectively, which can only be activated after certain specific instants; that is, the triggering condition needs to be monitored only after the event-triggering mechanism has been activated. The time interval between the event-triggered instant and a specific activation instant ensures that the event-triggering mechanisms for the detectors and the agents have PMIETs. Finally, the application of spacecraft formation tracking in low Earth orbit illustrates the availability of the theoretical results.

The leader–follower dynamic event-triggered controller is proposed for multiagent systems (MASs) subject to input time delay under time-varying coupling control gain. To begin with, the Halanay inequality, which is referred to as a delayed differential inequality, is used to deal with the input time-delay leader–follower problem. Second, a time-varying event-triggered control protocol with a dynamic triggering mechanism is adopted to save the communication resources with the satisfaction of state synchronization among agents at an exponential convergence rate. Additionally, a novel dynamic triggering mechanism that does not require continuous communication information among agents is further generated with more trigger times than before. Both dynamic triggering mechanisms show better superiority in excluding Zeno behavior and saving communication resources due to the embedding of a positive dynamic parameter in the triggering mechanisms. Finally, a simulation example and the F16 aircraft systems are presented to demonstrate the availability of the theoretical results.

In this article, subject to time-varying delay and uncertainties in dynamics, we propose a novel adaptive fixed-time control strategy for a class of nonlinear bilateral teleoperation systems. First, an adaptive control scheme is applied to estimate the upper bound of delay, which can resolve the predicament that delay has significant impacts on the stability of bilateral teleoperation systems. Then, radial basis function neural networks (RBFNNs) are utilized for estimating uncertainties in bilateral teleoperation systems, including dynamics, operator, and environmental models. Novel adaptation laws are introduced to address systems' uncertainties in the fixed-time convergence settings. Next, a novel adaptive fixed-time neural network control scheme is proposed. Based on the Lyapunov stability theory, the bilateral teleoperation systems are proved to be stable in fixed time. Finally, simulations and experiments are presented to verify the validity of the control algorithm.

This article focuses on flexible single-link manipulators (FSLMs) under boundary control and in-domain control. The actuators of the system include the dc motor at the end of the joint and m piezoelectric controllers installed at the flexible link, which is regarded as an Euler-Bernoulli beam. The problem of the infinite number of actuator failures, including the partial loss of the effectiveness and total loss of effectiveness, is solved by the adaptive compensation method. By introducing the relative threshold strategy, the event-triggered control (ETC) scheme is proposed to achieve angle regulation and vibration suppression while reducing the communication burden between the controllers and the actuators. The Lyapunov direct method is utilized to prove that the system is uniformly ultimately bounded and both the angular tracking error and elastic displacement converge to a neighborhood of zero. Numerical simulation results are provided to demonstrate the effectiveness of the proposed control law.

This article focuses on developing a static output feedback (SOF) control scheme for Takagi–Sugeno fuzzy systems. The proposed SOF controller does not share the same premise membership functions with the model, which permits enhancing the flexibility in controller design and implementation. By contrast with the state feedback case, the SOF control generally results in nonconvex design conditions. To circumvent this problem, we develop a successive convex optimization algorithm, which is based on solving a sequence of more tractable convex optimization problems obtained by approximating the nonconvex constraints with some convex ones. As a heuristic algorithm, the validity of the developed successive convex optimization algorithm is highly affected by initial conditions, and, recognizing this, we put forward an iterative procedure for determining the feasible initial condition. Finally, two illustrative examples are presented to validate the efficiency of the proposed algorithms.

This article discusses$H_{\infty }$fuzzy control for the nonlinear fourth-order parabolic equation with input delay via collocated actuator/sensor pairs. We suggest that the interval [0, 1] is divided into$M$subdomains, where sensors provide spatially averaged/point discrete-time state measurements. The control design strategy is proposed based on output measurements. We derive constructive conditions ensuring that the resulting closed-loop system is internally exponentially stable and has$H_{\infty }$performance by means of the Lyapunov approach.

A neural learning-based finite-time control policy is presented for a robotic manipulator with unknown backlash-like hysteresis and system uncertainties. Adaptive neural networks are adopted to deal with unknown robotic dynamics. In order to eliminate the effect of unknown backlash-like hysteresis, a robust adaptive term is designed in the backstepping design process. A neural network-based finite-time controller is designed by introducing a fractional order term, which guarantees the finite-time convergence of both neural networks and adaptive terms, and this type of convergence improves control accuracy to a certain extent. With the Lyapunov stability theory, the proposed scheme can be proved to make the errors be semiglobally finite-time stable (SGFTS). The effectiveness of the proposed control is shown by simulation and experimental results.

This article studies a single boundary regulator for the flexible tail of an autonomous robotic fish to implement complex oscillating body motions. The dynamic model of the flexible tail is derived by Hamilton’s principle, and conforms to the partial nonuniform Euler–Bernoulli beam with the uneven parameters. Then, a boundary control at the body–tail junction is proposed to manipulate the oscillation of the flexible tail, which is given in the form as a torque. The exponential stability of the error system is deduced by the integral Lyapunov synthesis. For further considering the boundary disturbance at the same point with control and the distributed disturbance, a disturbance observer is proposed. By appropriately choosing the designed parameters, the uniformly ultimate boundedness with disturbances is proved and the tracking error converges to a small neighborhood of 0. Finally, some simulations are presented to illustrate the effectiveness of the proposed control.

A finite-time control method is presented for n -link robots with actuator saturation under time-varying constraints of work space. Barrier Lyapunov functions (BLFs) are designed for ensuring that the robot remains under time-varying constraints of the work space. In order to deal with asymmetric saturation nonlinearity, we transform asymmetric saturation into a symmetric one by using a hyperbolic tangent function, which is introduced to avoid the discontinuous problem existing in the auxiliary system-based saturation method. Combining fuzzy-logic systems (FLSs) with the backstepping technique, a finite-time control policy is designed for ensuring the stability of the closed-loop system. With the use of the Lyapunov stability theory, all the error signals are proved to be semiglobal finite-time stable (SGFS). Finally, the experiment is carried out to verify the effectiveness of the finite-time method.

This article is concerned with the design of robust resilient control strategy for nonlinear systems under Denial-of-Service (DoS) attacks. First, Takagi–Sugeno fuzzy model is employed to approximate the nonlinear dynamics, and an improved sensor system is constructed by fuzzy observer and fuzzy predictor. By applying periodic event-triggered control strategy, the relationship between the estimated state and predicted state is obtained by event-triggering mechanism, which can reduce transmission attempts significantly in the sensor-to-controller channel. Second, caused by DoS attacks, the transmission attempts may be denied over the communication network. Under the proposed transmission policy with a bounded update period, input-to-state stability (ISS) of closed-loop systems can be guaranteed. Moreover, the maximum frequency and duration of DoS attacks are calculated. Finally, simulation results are provided to show the effectiveness of the proposed method.

A novel resilient structure is proposed to solve time-varying output formation tracking problem of heterogeneous multiagent systems (MASs) with unknown leader under contested environments in this article. The proposed novel resilient structure can not only attenuate the impact of attacks on the sensor and actuator of follower but also compensate the effect of attacks on the communication network. It contains two types of attacks in contested environments, one of which is an attack on the sensor or actuator of followers, called Attack type I. Another attack on communication networks is Attack type II. Solving the time-varying output formation tracking problem is divided into two steps. The first step is to design the compensation control signal to realize formation problem and the second step is to design the optimal control law to realize the tracking problem. Assuming that the state of leader is unknown, a novel resilient observer is first designed to estimate the state of leader for each follower, as well as compensate the effect of Attack type II. Then, with the information of observer,$H_{\infty }$controller is designed for each follower to attenuate the impact of Attack type I. Finally, the effectiveness of the proposed novel resilient structure is verified by a simulation example.

Gait recognition, applied to identify individual walking patterns in a long-distance, is one of the most promising video-based biometric technologies. At present, most gait recognition methods take the whole human body as a unit to establish the spatio-temporal representations. However, we have observed that different parts of human body possess evidently various visual appearances and movement patterns during walking. In the latest literature, employing partial features for human body description has been verified being beneficial to individual recognition. Taken above insights together, we assume that each part of human body needs its own spatio-temporal expression. Then, we propose a novel part-based model GaitPart and get two aspects effect of boosting the performance: On the one hand, Focal Convolution Layer, a new applying of convolution, is presented to enhance the fine-grained learning of the part-level spatial features. On the other hand, the Micro-motion Capture Module (MCM) is proposed and there are several parallel MCMs in the GaitPart corresponding to the pre-defined parts of the human body, respectively. It is worth mentioning that the MCM is a novel way of temporal modeling for gait task, which focuses on the short-range temporal features rather than the redundant long-range features for cycle gait. Experiments on two of the most popular public datasets, CASIA-B and OU-MVLP, richly exemplified that our method meets a new state-of-the-art on multiple standard benchmarks. The source code will be available on https://github.com/ChaoFan96/GaitPart.

The optimal solution to the leader-follower bipartite output synchronization problem is proposed for heterogeneous multiagent systems (MASs) over signed digraphs in the presence of adversarial inputs in this article. For the MASs, the dynamics and dimensions of the followers are different. Distributed observers are first designed to estimate the leader's two-way state and output over signed digraphs. Then, the leader-follower bipartite output synchronization problem on signed graphs is translated into a conventional output distributed leader-follower problem over nonnegative graphs after the state transformation by using the information of followers and observers. The effect of adversarial inputs in sensors or actuators of agents is mitigated by designing the resilient H∞controller. A data-based reinforcement learning (RL) algorithm is proposed to obtain the optimal control law, which implies that the dynamics of the followers is not required. Finally, a simulation example is given to verify the effectiveness of the proposed algorithm.

This paper is concerned with the dissipativity-preserving model reduction problem for Takagi–Sugeno (T–S) fuzzy systems. The principal goal is to approximate the high-order T–S model with a dissipative reduced-order T–S model. The number of fuzzy rules and the membership functions of the reduced-order T–S model are chosen freely to enhance design flexibility. To this end, an$H_{\infty }$performance index is used to describe the approximation error. Meanwhile, dissipativity of the reduced-order model is guaranteed by satisfying a dissipation inequality. With the aid of fuzzy-basis-dependent Lyapunov functions and slack variable techniques, less conservative design conditions for reduced-order models are derived. An algorithm is proposed to calculate a desired reduced-order model. A rail traction control system is given to illustrate the effectiveness of the proposed method and the advantages over the existing methods.