This article proposes a sampled-data event-triggered adaptive neural network (NN) control strategy to cope with the digital communication and attack compensation problems of networked systems with actuator attacks and exogenous disturbance. By combining event triggering state, parameter estimation signals, and disturbance observer, a novel digital state feedback controller is designed to reduce its updating frequency and compensate for the deliberate impact of unknown actuator attacks. Moreover, considering that the state is partially measurable, a novel observer-based digital controller is designed via a double-ended event-triggering mechanism (ETM). Then, two new Lyapunov functionals are created to analyze the system stability, and two design methods are given to solve the control gain. Finally, the feasibility and validity of the derived results are verified by a visual servo control system and an offshore structure system.

Soft cable-driven lower-body exosuits offer flexibility and low mass compared to rigid exosuits. Mass is further reduced in ankle-only exosuits, taking advantage of the fact that ankle plantar flexion contributes approximately half of positive mechanical power during normal human walking. In this paper we report two contributions to soft lower-body ankle-only exosuits. First, we describe the design of an exosuit with a single actuator that is time-shared between the two ankles. The elimination of an actuator lowers the total mass and cost of the exosuit by taking advantage of the time-separation between periods when each ankle contributes positive power during walking. Second, we describe a novel bio-inspired control strategy that adapts the impedance at the assisted ankle in response to unknown terrain, mimicking natural human ankle variable impedance, while simultaneously learning the dynamic parameters of the exosuit. Experiments show rapid adaptation and convergence to stable impedance and dynamic parameters while subjects walk on different types of terrain. Moreover, subjects' muscular effort, as measured by EMG sensors, is reduced relative to the cases of no exosuit or an unpowered exosuit. Note to Practitioners-Soft wearable robots (exosuits) offer significant advantages over rigid exoskeletons for locomotion assistance, including lighter weight, lower inertia, and improved wearability. Recognizing that ankle plantar flexion generates roughly half the positive power during walking, we developed a lightweight soft exosuit employing a single actuator to provide time-shared assistance to both ankles, reducing system mass and complexity. Especially, our bio-inspired adaptive control strategy enables the exosuit to learn and mimic the user's natural ankle impedance in real-time while walking over varying terrain. This biomimetic approach, where the exosuit adapts its interaction torque, impedance (stiffness/damping), and trajectory to match the user's inherent response to the ground, makes the assistance feel more intuitive and predictable, fostering user trust. Additionally, it exhibits an effective reduction in the user's muscle effort. This combination of comfort, adaptability, and trust-enhanced control improves work efficiency and quality for individuals requiring prolonged walking or standing support, such as the elderly, industrial workers, or soldiers, mitigating fatigue and potential injuries. Future work will explore optional actuators for specific scenarios and further enhance user comfort and trust through improved ergonomics and interaction transparency.

This article investigates the leader-following bipartite time-varying formation (BTVF) control problem for switched multiagent systems (MASs) under the changeable directed signed topologies. The communication topology switches to obey an average dwell-time condition, capturing realistic network dynamics. Two critical challenges are addressed in the controller design, one of which is that the real states are inaccessible and the other is that we know nothing about the active leader input. These constraints significantly increase the design complexity. Against this backdrop, we develop a novel control protocol that is capable of achieving the BTVF tracking even given the uncertain leader input without using the real states. The designed control protocol can perform well to deliver reliable commands to realize the BTVF under switched topologies. Through Lyapunov stability analysis and recursive algorithms, we rigorously prove convergence to the desired BTVF. Notably, the protocol is also shown to guarantee bipartite consensus as a special case. In the end, the effectiveness of the proposed control scheme is validated through simulations involving the clusters of the wheeled mobile robot and autonomous aerial vehicle models in different working scenarios.

The remaining useful life (RUL) prediction of ship electric propulsion system faces severe challenges such as multicondition switching and complex degradation modes. Under complex coupling conditions involving significant distribution differences across multisource domains and quantitative requirements for prediction uncertainties, conventional multisource methods struggle to resolve feature shifts caused by coupled condition-mode differences, as they primarily focus on single-aspect solutions. To this end, an RUL prediction method that integrates the entropic Gromov-Wasserstein (EGW) distance and adaptive cooperative intra-inter domain adversarial network (ACIDAN) is proposed in this article. First, a divergence measurement model for the EGW distance distribution is constructed, and the first predicted time is accurately identified through adaptive parameter optimization. Then, the ACIDAN is designed. Intradomain adversarial training is employed to eliminate feature-distribution shifts under multiple working conditions. The outputs of multisource regressors are coordinated through an interdomain kernel-density-estimation weighting strategy to ensure consistency. A multisource predictive variance method for the residual connected regressor is introduced in the degraded perception stage to generate confidence intervals containing uncertainty quantification. Finally, the RUL prediction accuracy and uncertainty quantification of the algorithm proposed in this article are verified.

This paper focuses on event-triggered practical prescribed-time consensus tracking problem of second-order multiagent system (MAS) with unknown nonlinear dynamics. The design consists of three major steps. Firstly, a new edge-based dynamic event-triggered prescribed-time distributed observer is constructed to estimate the leader’s states, which reduces communication frequency and response delay, requires no global information and obtains leader’s states within prescribed time without numerical implementation difficulty. Secondly, a novel relative threshold dynamic memory event-triggered prescribed-time control protocol is developed, which not only reduces control updating frequency and computation burden, but also improves control performance and achieves practical prescribed-time consensus tracking. Finally, stability analysis and non-Zeno behavior analysis are conducted, authenticating that the developed distributed observer and control protocol are practical prescribed-time stable without Zeno behavior. Simulation results also verify the effectiveness of the proposed method.

Recently, the operational domain of uncrewed aerial vehicles (UAVs) has expanded from outdoor environments to indoor spaces such as factories, power plants, and tunnels. In these environments, where the Global Navigation Satellite Systems (GNSS) is unavailable, and the risk of collision with surrounding structures is high, a positioning system with high accuracy and fast update rates is essential. In this article, we propose a novel 3-D positioning algorithm for UAVs operating in a GNSS-denied environment. The proposed algorithm is based on sensor fusion of the inertial measurement unit (IMU) and ultra-wideband-based wireless sensor network (UWB-WSN). The proposed algorithm, referred to as the recursive extended finite-memory positioning (REFMP), features a unique finite-memory (FM) structure that estimates the current position using only a limited set of recent information. This FM structure endows the algorithm with robustness against measurement model errors, linearization errors, and uncertainties in initial position information. We evaluated the performance of the proposed algorithm through both simulations and real-world experiments under various challenging conditions, including UAV collisions with walls, aggressive maneuvers inducing sensor measurement errors, and scenarios with uncertain initial position information. By comparing the proposed algorithm with state-of-the-art UAV positioning algorithms based on IMU and sensor fusion, we demonstrate the superior positioning accuracy and robustness of REFMP under harsh conditions.

This work presents a practical finite-time tracking control scheme for two-wheeled mobile robots in the absence of available reference linear and angular velocities. In contrast to the existing works, the proposed approach achieves the desired control performance by integrating practical finite-time stability theory into a backstepping framework, wherein a novel auxiliary variable is introduced. In addition, based on this auxiliary variable, the corresponding control laws are designed to achieve the tracking goal even if the bounds of the reference velocities are unknown. To highlight the importance of a disturbance-free network for transmitting reference velocities and the practical contributions of this work, a simulation result is presented to demonstrate the significant impact of such disturbances on tracking performance, whereas the effectiveness of the proposed tracking control laws for two-wheeled mobile robots is verified through experimental results.

The autonomous recovery of autonomous underwater vehicles (AUVs) using unmanned surface vehicles (USVs) is a critical prerequisite for achieving surface and underwater collaboration. The existing recovery methods primarily rely on the maneuverability of the AUV to complete the recovery, which significantly impacts the recovery success rate in complex marine environments. In this context, we propose a novel dynamic recovery control scheme based on the designed catamaran USV to recover the AUV actively. The features of the developed recovery controller are threefold: 1) by transforming the dynamic recovery control problem into a trajectory tracking control problem, a dynamic recovery approach is constructed; 2) a novel performance function is incorporated into the $\mathbf{tan}$ -type barrier function to provide the reasonable recovery position and attitude, which ensures that the AUV can be constrained within the range of the recovery cage; and 3) the issues of input saturation, initial speed jump, and external disturbances are addressed by employing auxiliary dynamic system, bioinspired neurodynamics, and disturbance observer, respectively. With the developed controller, the tracking errors can be guaranteed to converge into a small neighborhood of the zero. Finally, both simulation and experimental results are presented to illustrate the effectiveness of the developed recovery control algorithm.

This paper investigates distributed receding horizon estimation (DRHE) for time-invariant discrete-time linear systems over a sensor network. The original system is decomposed into several low-dimensional subsystems, where each sensor is capable of observing only one specific subsystem. The observable substate for each node is estimated by minimizing a local cost associated with receding horizon estimation (RHE), while the prediction of unobservable substates is updated through one-step weighted fusion. A maximal directed acyclic graph (MDAG) is introduced to facilitate the construction of the weight values for fusing these predictions, which is a more general method compared to directed spanning trees. Additionally, we propose a novel algorithm for identifying an MDAG across the network. We establish sufficient stability conditions for the proposed estimator under the assumption of collective observability. Finally, a numerical example of temperature monitoring is presented to demonstrate the effectiveness of the developed method.

This paper proposes a finite-time safety-critical formation control framework for multi-agent systems and validates it on a multi-unmanned aerial vehicles (UAVs). First, a finite-time formation controller based on a virtual leader-follower architecture is designed as a nominal controller to enable fast formation tracking and recovery. Second, a finite-time control barrier function quadratic programming (FT-CBF-QP)-based safety filter is added after the nominal controller to ensure inter-agent collision avoidance and obstacle avoidance under actuation limitations. Then, a symmetry-breaking mechanism is incorporated to mitigate safe-but-stagnant deadlock under dense interactions. Finally, the framework is applied to Crazyflie UAVs, and simulation results on Robot Operating System 2 demonstrate faster formation tracking than conventional methods and reliable collision-free behavior with fast post-avoidance convergence.

In this article, we propose a new finite memory-based sliding mode control (FM-SMC) for robust quadcopter trajectory tracking. The new FM-SMC was developed by designing a finite memory-based disturbance observer (FM-DOB) for accurate and robust disturbance compensation and a finite memory-based neural network learning algorithm (FM-NNLA) for approximating unknown nonlinearities. Unlike conventional infinite-memory approaches that suffer from error accumulation and sensitivity to initial conditions, the proposed FM-DOB estimates and compensates for disturbances within a finite time horizon, while the FM-NNLA updates neural network weights using only recent state information to approximate nonlinear dynamics without long-term error accumulation. By leveraging these FM structures, the new FM-SMC enhances robustness against disturbances and system uncertainties while ensuring stable performance. Rigorous stability analysis is carried out using Lyapunov theory, and real-time quadcopter experiments on boustrophedon and ascending helical trajectories verify the robustness and superior effectiveness of the new FM-SMC.

This paper introduces a decentralized control framework for modular aerial delivery systems, leveraging a Graph Neural Network-based controller to enable robust and adaptive flight across various parcel configurations. The system comprises homogeneous propeller modules that are flexibly attached around a parcel, with each module relying only on local state and 1-hop neighbor information communicated over a robust wired network. The proposed controller is trained offline to emulate the thrust differentials of an optimal centralized controller, enabling stable flight without requiring global state estimation or centralized coordination. Experimental evaluations on two parcel types (0.7 kg and 1.7 kg) with quadrotor, hexacopter, and octocopter configurations demonstrate reliable performance: hover tests achieve roll, pitch, yaw, and altitude standard deviations below 0.3 degrees, 0.3 degrees, 0.2 degrees, and 0.02 m, respectively; L-shaped trajectory tracking shows velocity errors below 0.1 m/s and path-following errors under 0.05 m. The predicted thrust norms exhibit mean squared error deviations of only 5%-10% relative to the optimal controller across all configurations, underscoring the high fidelity of the learned control policy. The proposed framework provides a scalable and robust solution for decentralized aerial delivery, addressing practical challenges in center-of-mass variation and asymmetric payload arrangements.

This paper investigates the issue of adaptive neural non-fragile proportional and derivative (PD) feedback control for the singular systems with unknown nonlinear dynamics. First, considering the inaccuracy of controller implementation, the problem of non-fragile controller design is considered and solved by using a robust control strategy. Second, PD feedback control is established to transform the singular system into a normal system, which facilitates stability analysis of the system. Third, the adaptive proportional-derivative radial basis function neural network technique is used to approximate the unknown nonlinear function and resist its influence. Under this designed framework, the stability conditions of the closed-loop system are given by using the Lyapunov method. The designed methods of state feedback gains and observer-based gain matrices are presented, respectively. Last, three examples are employed to elucidate the feasibility of the developed control strategy.

To address the problem of permanent magnet linear synchronous motor (PMLSM) velocity tracking in high-sampling-rate systems, this paper focuses on achieving high tracking accuracy and disturbance rejection capability. A delta operator framework is employed to establish a consistent description between continuous and discrete domains, unifying PMLSM velocity controller design. First, within this framework, a data-driven ultra-local model (ULM) is utilized to reduce the reliance of traditional dual-loop control systems on precise mathematical models. A dual finite-time disturbance observer is designed to estimate the complex disturbances and uncertainties inherent in the ULM. Furthermore, by incorporating a prescribed performance control (PPC) function, a modified integral sliding mode controller is developed to suppress velocity chattering and guarantee that the tracking error remains within specified bounds. Finally, this paper establishes a theoretical connection between continuous and discrete systems under the delta operator framework and provides experimental validation. Simulations and experiments on the RTU-BOX PMLSM platform demonstrate improved dynamic tracking performance at high sampling rates and verify the practical feasibility of the proposed method.

An interval type-2 fuzzy system (IT2FS)-based actuator failure compensation strategy is introduced for nonlinear systems including uncertain dynamics, unknown control direction and actuator failure. The proposed compensation scheme deals with lock-in-place and total or partial loss of effectiveness failures, simultaneously and uses IT2FS to model uncertain functions and uncertain changes obtained due to the actuator failures. Also, to handle the unknown control direction arising because of the unknown control coefficient and uncertain loss of effectiveness failure, the Nussbaum function idea is invoked. After that, the proposed scheme is designed using adaptive command filtered backstepping (CFB) approach. The suggested scheme eliminates the explosion of terms as well as difficulties in selecting the time constant of filters in the dynamic surface control (DSC) and compensates filtering errors. The presented compensation approach assures boundedness of the closed-loop signals. Finally, simulation and comparison results on two different applicable examples under different actuator failures are presented to highlight the effectiveness and performance of the presented approach.

The task performance of mobile manipulators can be significantly enhanced by whole-body control and optimization in complex scenarios. Due to the nonlinear properties of whole-body dynamics and parameter uncertainty, modeling accurate system dynamics is essential in addition to designing an effective control strategy. However, traditional control methods have high computational costs and fail to deal with the parameter errors caused by model linearization. To address these issues, we propose a model predictive control (MPC)-Net, a learning-based approach that facilitates rapid online optimization by combining deep learning with multiple MPCs. Firstly, we develop a parameter identification algorithm based on a deep neural model to estimate the unknown dynamics parameters. Although the control performance of the MPC approach positively correlated with the prediction horizon, a long horizon would result in additional computational costs. Thus, MPC-Net is constructed by combining multiple sub-MPC issues, and the nonlinear coefficients are obtained by using a deep neural network-based optimization framework. Furthermore, MPC-Net generates the solution by combining the outputs of multiple sub-MPC problems using the nonlinear transformation of learned coefficients. Experiments are conducted on a mobile manipulator, which demonstrates the proposed MPC-Net-based optimization control offers fast efficient computation and low tracking error performance.

This work introduces a fixed-time adaptive fault-tolerant control approach to tackle the trajectory tracking issues in wheeled mobile robots by accounting for unknown dead zones and actuator faults. Firstly, by introducing smoothing functions, bounded estimates, and adaptive parameters, combined with a fixed-time control strategy, actuator faults and the impacts of unknown dead zones are effectively eliminated. Second, an adaptive fault-tolerant control strategy for nonlinear wheeled mobile robot systems is developed. Subsequently, a fixed-time trajectory tracking controller is designed to guarantee system stability and achieve high-precision control for wheeled mobile robots. Furthermore, leveraging the principles of Lyapunov stability theory, this work rigorously establishes the convergence properties of the controller, offering a robust theoretical foundation for ensuring the precision and reliability of trajectory tracking. Finally, the proposed control strategy's effectiveness and feasibility are validated through simulation outcomes. Note to Practitioners-With the rapid development of the field of intelligent control, the trajectory tracking control of wheeled mobile robots is also getting more and more attention. However, in practical applications, wheeled mobile robots are not only subject to external disturbances, but also internal disturbances exist. For the external interference, we have already had more research, while the internal interference like unknown dead zone and actuator failure has not been considered enough in the practical application. In order to solve this problem, the bounded estimation and adaptive parameters are used to compensate for the influence of unknown dead zone and actuator failure. In addition, considering the timeliness in practical applications, the corresponding controller is further designed based on the fixed-time control law. Finally, the effectiveness of the Proposed control program is illustrated by a simulation example.

Cyber-physical systems (CPS) are everywhere around us-from smart grids that power our cities to intelligent transportation systems that keep traffic moving. These systems work by tightly combining software, communication networks, and physical equipment. But that same tight integration comes with a downside: it also creates new opportunities for cyberattacks. And when CPS are attacked, the consequences can be far more serious than in traditional IT systems-impacting not only money and infrastructure, but potentially public safety as well. In this talk, I will look at CPS security from an attacker's point of view. I will begin with a brief introduction to what CPS are, why they are vulnerable, and what kinds of cyber-attacks have already been studied and observed. I will then share our recent work on stealthy hybrid attacks-a class of attacks that combine multiple cyber techniques to cause maximum disruption while staying as hidden as possible. These attacks are designed to degrade system performance significantly, yet avoid triggering existing detection mechanisms. By understanding how such attacks can be constructed, we can better identify the weaknesses of current defense methods and, more importantly, develop new strategies for detection, resilience, and secure operation. The goal is to help future CPS remain not only smart and efficient, but also safe, reliable, and trustworthy.

Accurate apple detection and precise three-dimensional (3D) localisation are essential for autonomous robotic harvesting in orchard environments, where occlusion, illumination variation, depth noise, and the similar colour appearance of fruits and surrounding leaves present significant challenges. This paper proposes a dual-detector vision framework combined with depth-aware back-projection to achieve robust apple detection and metric 3D localisation in real time. The method integrates the complementary strengths of YOLOv8 and Mask R-CNN through confidence-weighted fusion of bounding boxes and pixel-wise union of segmentation masks, producing stabilised two-dimensional (2D) apple representations under visually ambiguous conditions. The fusion results are converted into dense 3D representations through depth-guided projection within the camera coordinate system representing the visible fruit surface. A depth-consistency weighting strategy assigns higher influence to depth-reliable pixels during centroid computation, thereby suppressing noisy or occluded depth measurements and improving the stability of 3D fruit centre estimation, while local intensity normalisation standardises neighbourhood-level pixel intensities to reduce the impact of shadows, highlights, and uneven lighting, enabling more consistent segmentation and detection across varying illumination conditions. Experimental results demonstrate an accuracy of 98.9%, an mAP of 94.2%, an F1-score of 93.3%, and a recall of 92.8%, while achieving real-time performance at 86.42 FPS, confirming the suitability of the proposed method for robotic harvesting in challenging orchard environments.