Keywords: Fault Detection and Diagnostics, Motion Control, Control Applications
Abstract: To ensure both recovery performance and convergence time, this paper proposes a novel incremental Fault-Tolerant Control (FTC) integrated with an event-triggered mechanism. A function with prescribed performance is designed to constrain the tracking errors, guaranteeing that they remain within the prescribed safety boundary at all times. By incorporating the prescribed-time control theory into incremental control, the tracking errors recover to a stable state within a user-defined time. To reduce computational burden, an event-triggered mechanism is further integrated to the proposed prescribed-performance and prescribed-time incremental FTC to avoid unnecessary control updates. Finally, based on the proposed incremental FTC, the angular rate controller for fixed-wing aircraft is developed to improve flight safety. Simulation results demonstrate that the developed FTC effectively overcomes the asymmetric wing damage, achieving the desired recovery performance and convergence time while significantly reducing computational cost.

Keywords: Multi-agent Systems, Control Applications, Automated Guided Vehicles
Abstract: This paper investigates the lag consensus problem arising from leader state information transmission in multiunmanned aerial vehicle (multi-UAV) leader-follower communication network and develops a prescribed time (PT) lag consensus protocol. The stability of the control law is rigorously established through construction of an appropriate Lyapunov function, accompanied by derivation of the corresponding stability criteria. To enhance system’s performance, an adaptive weight pigeon-inspired optimization (AWPIO) is developed, incorporating an adaptive weight adjustment mechanism for controller parameter tuning and energy consumption minimization. The efficiency of the proposed control protocol and optimization algorithm is comprehensively demonstrated through numerical simulation and comparative experiment.

Keywords: Robotics, Real-time Systems, Learning Systems
Abstract: This work proposes a bimanual force-feedback teleoperation platform designed for collecting and learning complex manipulation tasks. The system integrates two Franka robotic arms as the leader and another two as the follower, supporting real-time force feedback and high-precision multimodal data collection. To enhance the performance of diffusion-style policies, several algorithmic improvements are introduced, including an efficient 3D point cloud encoder, a stochastic interpolation framework to reduce distribution discrepancies, and an inpainting-based method to improve action continuity. Additionally, large language models (LLMs) are incorporated to enable natural language interaction, task instruction parsing, and semantic execution. We validate the system in a “congee shop” scenario, where it autonomously performs tasks such as adding ingredient, water pouring, cooking, serving, and so on. Experimental results demonstrate the system's effectiveness in handling multimodal, contact-rich tasks and interacting naturally with users.

Keywords: Linear Systems, Optimal Control, Learning-based Control
Abstract: This paper considers a two-player linear quadratic differential hypergame where Player 2 holds misperception about the objective of Player 1. Such a problem arises in practical situations such as mixed human-autonomous driving, where the autonomous vehicle may misinterpret human driving intentions. Such misperception typically precludes exact hyper-Nash equilibria and brings technical challenge in the design of the optimal strategy for Player 2 due to obscured Nash-relevant parameters under state-only observations. To address the difficulty, we develop an inverse learning-based method that reconstructs the Nash-relevant closed-loop dynamics induced by the opponent's strategy from finite state trajectories. Based on the recovered game structure, a Riccati flow-based strategy update law is then designed, which drives the proposed strategy toward the exact Nash equilibrium of the underlying game. The effectiveness of the proposed strategy is validated by a car-following case with misperception between an autonomous vehicle and a human driver.

Keywords: Automated Guided Vehicles, Learning-based Control, Robotics
Abstract: The Reinforcement Learning (RL) provides an effective paradigm for control strategy design of autonomous vehicles (AV). However, the actual application of RL is hindered by challenges of training inefficiency and runtime instability. This paper proposes a novel RL framework with online MPC augmentation for path tracking. To enhance the online control performance of the RL, the proposed framework introduces an MPC controller to provide real-time constrained augmentation to the control signal of the RL controller. In addition, the augmented control and the magnitude of MPC compensations are fed back to guide the update of state and rewards during the retraining of RL. In this manner, the AV agent explicitly learns the interaction between learning-based control and optimization-based augmentation, enabling faster convergence of the training stage. Simulation results demonstrate that the proposed framework achieves competitive tracking control accuracy, while drastically reducing the RL training overhead with a large magnitude of reduction in convergence time.

Keywords: Multi-agent Systems, Robotics, Adaptive Control
Abstract: UAV swarms tracking moving targets in cluttered environments face coupled challenges: bearing-only observability limits, dynamic uncertainty diffusion, and environmental occlusions. To address these, we propose a completely distributed, unified active perception and planning framework. First, a Distributed Recursive Least Squares (DRLS)-based estimator circumvents traditional linearization errors to guarantee state convergence under bearing-only constraints. Second, to proactively suppress uncertainty, an entropy-driven formation strategy leverages the kinematic evolution of the error covariance, steering the swarm to maximize information gain against the principal uncertainty axes. Third, bridging active guidance with safe execution, a kinodynamic planner generates collision-free trajectories that explicitly resolve Field-of-View (FoV) limits and teammate occlusions. Extensive high-fidelity simulations demonstrate that the proposed system ensures persistent tracking in dense clutter, significantly reducing theoretical uncertainty and estimation errors while maintaining exceptional formation survivability and flexibility.

Keywords: Nonlinear Systems and Control, Robotics, Motion Control
Abstract: Jumping robots, which offer high energy density and tunable properties, exhibit superior obstacle-crossing capability for exploration missions. However, most existing studies rely on simplified linear spring assumptions for modeling, which fail to accurately capture the nonlinear mechanical behavior of large-deformation composite leaf springs. To address this issue, this paper presents an equivalent mechanical model of a carbon fiber-reinforced polymer (CFRP)-based jumping mechanism derived from geometrically nonlinear theory. First, for a rec-tangular CFRP leaf spring compressed at both ends, a circular arc assumption is introduced to describe large-deflection de-formation. Using the variational principle, an analytical rela-tionship between compression displacement and elastic force is derived in the form of elliptic integrals. Second, an experimental platform consisting of a servo motor, reduction gears, a winding roller, and sensors is developed to enable high-precision compression loading via closed-loop position proportional-integral-derivative (PID) control. Mechanical tests are conducted on CFRP leaf springs of various specifica-tions. Finally, an empirical correction coefficient is introduced to calibrate the parameters of the theoretical model. Experi-mental results show that the calibrated model achieves a coef-ficient of determination R² above 0.99 and a root mean square error below 5% of the peak force, validating its predictive accuracy within a compression range of less than L0/2. The proposed mechanical model provides a reliable theoretical basis for the optimal design and performance prediction of elastic elements in jumping robots.

Keywords: Real-time Systems, Signal Processing
Abstract: 以应对快速计算中的关键挑战 复合体内的关键性能参数 在现代电网的环境下，本文提出了一个 现代电网快速计算方法与系统 性能参数。该系统利用稀疏 以频域能量映射架构为核心， 结合可重构的动态观察窗口 机制。它从根本上优化了计算 通过点状变换传统时域来建模 将密集运算转化为高效的频域标量 投射。实验结果与硬件部署 证明系统显著减少资源 在保持高负载的同时，消耗和处理延迟 精度，且监控性能符合 国际标准如IEC 61000-4-30，提供以下内容 一种高效且可重构的系统级解决方案 精细化的实时电网监测。

Keywords: Robotics
Abstract: Continuum and serpentine manipulators have attracted sustained interest due to their ability to navigate confined environments and safely interact with unstructured surroundings. Among existing modeling paradigms, constant-curvature representations offer an appealing trade-off between model fidelity and computational efficiency, enabling real-time control and planning. However, achieving reliable constant-curvature deformation in continuum manipulators remains challenging, particularly under external loading and practical actuation constraints. To address these challenges, this work introduces a novel continuum manipulator design that enforces constant-curvature deformation through mechanical synchronization. The core mechanism forces all discrete segments within a bending section to undergo equal rotation angles via a rigid coupling system.

Keywords: Robotics
Abstract: Abstract— Dense 3D mapping is fundamental to spatial perception and navigation in mobile robotics. Recently, the Visual Geometry Grounded Transformer (VGGT) has emerged as a powerful approach, offering rich dense geometric priors and fast inference speeds, with the ability to jointly process multiple images in a single forward pass in under one second. However, its limited input frame capacity constrains practical deployment in real-world scenarios. While RGB-D Visual Odometry (VO) provides reliable long-trajectory pose estimation, directly accumulating VGGT point clouds often leads to geometric distortions and global inconsistencies due to sensor noise and accumulated drift. To address these challenges, we propose a two-stage mapping framework that systematically couples dynamic RGB-D VO with VGGT point cloud processing to achieve high-fidelity 3D mapping. In the first stage, pose fusion is performed, wherein VO trajectories are leveraged to robustly initialize and refine the global poses of independent mapping sessions, ensuring local geometric stability. In the second stage, point cloud fusion is executed by aligning and merging VGGT point clouds under global optimization constraints to eliminate structural divergence. Experimental results demonstrate that our framework effectively enables long-trajectory dense mapping while significantly enhancing the global consistency of the reconstructed maps. By explicitly integrating reliable dynamic VO priors with VGGT-based point cloud refinement, our system delivers robust, high-quality static dense mapping in dynamic environments, achieving superior performance over existing RGB-D SLAM baselines. I. INTRODUC

Keywords: Robotics, Energy Efficiency
Abstract: This paper experimentally investigates the influence of passive joint configuration on the locomotion performance of a snake-like robot. The four-joint planar robot is used to evaluate multiple joint configurations, including fully actuated, three actuated joints with one passive joint (3A+1P), two actuated joints with two passive joints (2A+2P), and one actuated joint with three passive joints (1A+3P), under identical control inputs and physical conditions. Locomotion performance is quantitatively assessed using average forward velocity and cost of transport (COT). Experimental results show that incorporating passive joints can improve locomotion efficiency compared with the fully actuated configuration. Furthermore, different passive joint configurations exhibit distinct characteristics: the 3A+1P configuration achieves the highest forward speed under appropriate spring conditions, while the 2A+2P configuration demonstrates lower COT and better energy efficiency. These results indicate that locomotion performance strongly depends on passive joint placement, offering insights for the design of energy-efficient snake robots.

Keywords: Robotics, Estimation and Identification, Modeling and Control of Complex Systems
Abstract: Parameter identification for hybrid bipedal robotic legs remains challenging because parallel mechanisms introduce strong dynamic coupling and nonlinear friction is difficult to model accurately. This paper presents an enhanced identification method for a six-degree-of-freedom hybrid robotic leg with a 3-DOF serial hip, a serial knee, and a 2-DOF parallel ankle. The closed-loop kinematics of the parallel ankle are formulated, and a motor-to-joint torque mapping is derived using the principle of virtual work. The hybrid mechanism is then transformed into an equivalent serial multibody system for Lagrangian dynamic modeling. To improve model accuracy, motor current offsets are incorporated into the identification process to compensate for zero drift in low-torque regions. In addition, a trajectory optimization criterion formulated as a mass-weighted sum of the condition numbers of link-wise sub-regressor matrices is introduced to improve the balance of parameter excitation, subject to nonsingularity and excitation constraints. Experiments on the physical robotic leg show that current-offset compensation provides modest improvements under the baseline objective, while the final configuration combining the unified objective with current-offset compensation achieves the best overall performance. Compared with the baseline configuration, the final configuration reduces the NRMSE of Joint 1 and Joint 6 by 53.3% and 40.1%, respectively.

Keywords: Control of Smart Power Delivery Systems, Control of Distributed Generation Systems
Abstract: To accommodate the diversity and time-varying characteristics of grid regulation demands and flexible resource, efficient aggregation mechanisms are required to support grid regulation services. Dynamic aggregation adaptively adjusts resource composition according to grid demands, overcoming the limitations of the traditional fixed resource composition in virtual power plants (VPPs). However, the computational complexity of the large-scale resource composition is a new bottleneck. This paper introduces a dynamic aggregation optimization framework based on submodularity, which includes resource selection and coordination. A lazy submodular optimization method is proposed in the resource selection process to enhance the efficiency of flexible resource aggregation. Using the diminishing-return property of submodular functions, the proposed method caches previous aggregation gains as upper bounds, significantly reducing redundant evaluations. The method greatly improves computational efficiency while maintaining optimal aggregation quality. Simulations highlight the efficacy of the proposed method in dynamic aggregation.

Keywords: Energy Efficiency, Optimal Control, Automated Guided Vehicles
Abstract: To maintain efficient, economical driving for hybrid electric trucks (HETs), energy-saving cruise control must be used in complex dynamic traffic scenarios to optimize vehicle speed profiles and avoid unnecessary acceleration or deceleration. This paper proposes a Heuristic Dynamic Programming (HDP)- based energy-saving cruise control strategy enhanced with PINN. By incorporating power balance, SOC dynamics, and energy conservation as regularization terms into the loss function, PINN improves the model's predictive accuracy for complex nonlinear systems under physical constraints. The energy consumption model's output is incorporated into cruise control as an economic indicator for cruise-speed planning. Concurrently, an Informer neural network predicts the preceding vehicle's speed from historical data, mitigating the impact of the vehicle's uncertain driving behavior on energy-saving cruise control. HDP achieves optimal speed-cruise control in dynamic traffic scenarios by integrating the predicted speed of the preceding vehicle, the vehicle's predicted energy consumption, and its own state.

Keywords: Optimal Control
Abstract: This paper develops an efficient numerical algorithm to solve a class of discrete-time nonlinear optimal control problems with control constraints. Firstly, by introducing a virtual control variable, the original constrained optimal control problem is transformed into an unconstrained form. Subsequently, an Optimal Control Principle (OCP)-based algorithm with superlinear convergence is presented to solve the transformed problem. The computation of the gradient and Hessian matrix is further reformulated as optimal control problems of different types, and explicit forward-backward recursive formulas are derived using variational methods. Finally, numerical simulations are conducted to demonstrate the effectiveness of the proposed algorithm and its superior computational efficiency.

Keywords: Optimal Control, Linear Systems, Networked Control
Abstract: This paper proposes a DC (Difference-of-Convex) programming approach for designing contracting firing-rate neural networks. Based on non-Euclidean log-norm contraction conditions, we formulate a tunable optimization framework that jointly optimizes network parameters and metric weights. The resulting nonconvex design problem is reformulated as a standard DC program via a posynomial representation and logarithmic transformation. Numerical simulations show that the proposed method improves contraction margins and robustness-performance trade-offs compared with baseline designs.

Keywords: Optimal Control, Nonlinear Systems and Control, Motion Control
Abstract: Trajectory planning for heavy-duty trucks involves a critical trade-off between computational efficiency and dynamic fidelity, especially during high-speed emergency maneuvers. Standard kinematic planners cannot ensure safety under limit handling conditions, while high-fidelity nonlinear optimization methods struggle to meet the requirements of real-time feasibility. We propose Safety-Aware Augmented Lagrangian iLQR (SA-AL-iLQR) to bridge this gap. Our method combines a differentiable nonlinear tire model and actuator dynamics to accurately predict vehicle behavior. In addition, the hard safety constraints, including collision avoidance and rollover prevention, are reformulated as continuously differentiable functions. This enables SA-AL-iLQR to rigorously enforce safety boundaries, ensuring the Load Transfer Ratio (LTR) remains within limits. TruckSim simulations have verified that the planner can prevent instability when the kinematic baseline fails, demonstrating significant improvements in stability while meeting the 10 Hz real-time requirements on standard hardware, with negligible efficiency loss.

Keywords: Optimal Control, Nonlinear Systems and Control, Robotics
Abstract: Visual servoing (VS) can enhance the control precision of robotic manipulators by incorporating visual feedback into the closed-loop control process. However, in practical VS systems, it is challenging to ensure that the target object stays within the camera’s field of view while maintaining robustness against external disturbances. This work develops a robust model predictive control (RMPC) framework for hybrid VS (HVS) of robotic manipulators. In the proposed HVS scheme, image moments are selected as visual features to regulate the translational motion of the camera, whereas Euler angles are chosen to characterize the camera attitude for rotational regulation. A virtual camera model is further incorporated to decouple the image-moment kinematics from the camera rotation. Based on this decoupled model, we formulate an RMPC scheme with tightened state constraints that enforce state and input constraints under bounded disturbances. Moreover, sufficient conditions that ensure recursive feasibility and guarantee closedloop stability are rigorously established. Finally, simulation results demonstrate the robustness and constraint-handling capability of the proposed RMPC scheme.

Keywords: Estimation and Identification, Fault Detection and Diagnostics, Optimal Control
Abstract: Airborne inertial stabilized platforms in staring mode are vulnerable to false data injection attacks. In this mode,attack biases and physical disturbances can appear as similar slow drifts in the actuation channel; strong nonlinear dynamics and observability singularities further obscure their source. To overcome this attack-disturbance separation difficulty, we construct an observability-based attack-disturbance separability metric and propose an information-regularized model predictive control (IR-MPC) method. The method mimics raptor peering, where small deliberate head motions acquire additional visual information without breaking sustained gaze, to generate safE active excitation trajectories. The induced excitation enhances the separability between composite physical disturbances and attack signals, thereby supporting online attack identification, separation, and compensation. Simulations calibrated by prototype identification experiments show attack/friction convergence within 0.29 s/0.48 s after excitation starts, while keeping the excitation trajectory inside the FOV safety corridor

Keywords: Multi-agent Systems
Abstract: 本文提出了一个多队阵型问题，满足条件 时间窗口约束，旨在从中选择代理 候选人将组建多个团队以完成 多个任务，时间窗口不同。一个被选中的代理人 只能参与不相交的任务。此外，代理 任务执行可能失败。该问题范围广泛 申请的数量。数学模型被表述为 这个问题，以及一系列基于 提出了不同的启发式规则。计算 实验表明，其中三种算法 在两种解法方面表现出优异的性能 质量和计算时间。 索引术语——多代理团队组建、时间窗口 约束、贪婪算法。

Keywords: Multi-agent Systems, Adaptive Control, Learning-based Control
Abstract: This extended abstract proposes a distributed consensus tracking protocol for heterogeneous multi-agent systems with unmodeled nonlinear dynamics. To estimate unmeasurable leader states without dimensional constraints, a generalized distributed adaptive observer relying on leader observability is developed. Furthermore, fully adaptive matrix-form update laws are designed for both observation and feedback gains. This eliminates predetermined matrix reliance and resolves structural coupling via dynamic error-driven adjustments. Finally, a continuous robust control term with a dynamically decaying parameter smoothly compensates for uncertainties, ensuring chattering-free asymptotic stability. Simulations verify the proposed protocol.

Keywords: Multi-agent Systems, Adaptive Control, Robotics
Abstract: This paper investigates an adaptive Lyapunov-based distributed control for safe motion planning and leader-follower formation in multi-agent systems. The objective is to maintain a stable formation, allow additional agents to safely join the group, and guide all agents to designated goal positions. A distributed control law is derived, where each follower relies only on local information from the leader. Safety during motion and agent insertion is ensured through repulsive potential terms that enforce collision avoidance, and speed constraints. The proposed method guarantees formation stability and asymptotic convergence to the target positions. Simulation results with multiple robots demonstrate safe agent insertion, formation maintenance, and coordinated parking at goal locations.

Keywords: Multi-agent Systems, Learning Systems, Optimal Control
Abstract: Dual-mode heterogeneous systems integrating Power Line Communication (PLC) and Wireless Communication (WLC) are vital for ensuring reliability in the Power Internet of Things (PIoT). However, rule-based schemes lack adaptive collaboration among multiple Stations (STAs). Applying Multi-Agent Reinforcement Learning (MARL) to such systems faces two challenges: invalid action spaces from asynchronous channel states and reward misalignment due to heterogeneous transmission rates. To address these, this paper proposes Action Masking and Asynchronous Experience Replay for Dual-Mode Channel Access (AMA-DCA). Simulation results show that AMA-DCA significantly outperforms the Multiplexing algorithm in throughput, collision probability, and mean delay.

Keywords: Multi-agent Systems, Linear Systems
Abstract: This paper addresses leader--follower formation tracking for double-integrator agents while enforcing collision avoidance with other agents and static circular obstacles in the presence of stochastic acceleration disturbances. We design a distributed nominal formation controller and augment it with a risk-aware safety filter that embeds conditional value-at-risk (CVaR) into a control barrier function (CBF) condition. We use velocity-augmented distance barrier functions to reduce the relative degree to one and derive an affine-in-control form of the resulting CVaR--CBF constraints. This derivation yields a convex online quadratic program that minimally modifies the nominal input. Under a leader-rooted uniform joint connectivity condition, we prove asymptotic convergence of the disturbance-free formation error and provide a simple bound on the augmentation parameter that implies standard position-clearance constraints. Simulations illustrate safe obstacle avoidance and formation recovery under the proposed controller.

Keywords: Multi-agent Systems, Modeling and Control of Complex Systems, Robotics
Abstract: Resilient control of uncrewed aerial vehicle (UAV)swarm with attack containment strategy under denial-of-service (DoS) attacks is investigated in this paper. Inspired bystarling ffocks that achieve emergent global resilience through dynamic local topology adjustments, an attack containment strategy is proposed to identify and actively isolate disconnected nodes in ffnite time, preventing erroneous information propagation. Based on the connectivity identiffcation, a ffxed-time distributed observer is designed to ensure accurate leader state estimation.On this basis, a distributed resilient control framework is constructed, ensuring resilient swarm coordination under DoS attacks. The effectiveness of the proposed method is validated through theoretical analysis and numerical simulations

Keywords: Multi-agent Systems, Control Applications, Motion Control
Abstract: In this paper, a disributed control framework of multirobot system will be described in which a SIMATIC S7-1500 PLC (programmable logic controller) is employed to provide a stable source of sequence control, interlocks and safety logic, and MATLAB is employed to provide a higher level of coordination, monitoring and processing of motion feedback. The process of data exchange in both directions between the PLC and MATLAB is done through OPC UA and operator control is done using WinCC HMI/SCADA with clear modes, alarms and diagnostics. Stable communication, proper work of the process sequence, and online reception of the results of movement of the robot during operation were experimentally tested and confirmed. The proposed solution helps to transform a set of manipulators into an operator-friendly industrial automation system with a focus on any high-tech system.

Keywords: Intelligent and AI Based Control, Learning Systems, Signal Processing
Abstract: 无人机图像语义分割支撑了低空经济的扩展，具有巨大的实际应用价值和广泛的发展前景，但其进展受限于稀缺的像素级注释数据集。数据增强是解决这一瓶颈的核心方案。尽管现有的数据增强方法在一定程度上提升了模型的鲁棒性，但仍存在显著的局限性。传统方法仅实现表面数据变异，未能引入新的语义结构，而生成方法依赖单一先验，导致生成的无人机场景中结构准确性差和语义错位。为此，我们提出了一种基于可控扩散模型的方法，融合像素级语义掩模、HED边缘映射作为视觉先验，以及通过自定义控制适配器提供语义指导的类别丰富文本提示。具体来说ʌ

Keywords: Intelligent and AI Based Control, Man-machine Interactions, Optimal Control
Abstract: As a basic example for realizing personal AGI, the author has formulated a simple problem of dealing with highway congestion and examined the problem-solving (decision-making) process. In this paper, for the example of determining the optimal sequence of simple actions, such as where to consider routes, whether to make detours, and which exit interchange to choose, shown is the results of optimal behavior control that encompasses recognition and planning based on a system model that includes personal preferences as part of situation and environment. The preferences that should be incorporated for personalizing AGI are considered and their relationship with the behavioral evaluation function is investigated. It is also shown that the evaluation function changes as the preferences of the actor (pAGI user) change, and as a result, the optimal behavior also changes in a complex manner.

Keywords: Intelligent and AI Based Control, Sensor/Data Fusion, Signal Processing
Abstract: Brain–computer interface systems based on motor imagery have attracted considerable attention because they allow users to control external devices without actual muscle movement. At the same time, reliable classification of motor imagery EEG signals remains difficult because these signals are highly variable across individuals and often contain noise and artifacts. In this study, we examined the possibility of subject-independent motor imagery classification using a compact deep learning model suitable for portable BCI applications. EEG data from the PhysioNet Motor Movement/Imagery database were used, and the analysis focused on left-hand and right-hand imagery tasks. To make the system more practical for wearable use, only six electrodes located over the sensorimotor cortex were selected. The EEG signals were preprocessed through filtering, artifact reduction, epoch segmentation, and normalization. The proposed EEGNet-Lite model was then compared with conventional CSP-based approaches combined with LDA and SVM classifiers. Performance was evaluated using a leave-one-subject-out cross-validation scheme with accuracy and F1-score as the main metrics. The results showed that EEGNet-Lite achieved the best overall performance, reaching 84.6% accuracy and an F1-score of 0.83. These findings suggest that a lightweight neural network can provide reliable subject-independent motor imagery classification even with a reduced number of EEG channels, which is important for the development of portable and wearable BCI systems

Keywords: Learning Systems, Estimation and Identification, Real-time Systems
Abstract: This paper examines digital readiness in elderly care systems through a comparative study of social service providers in Slovenia and Croatia within the framework of smart age-friendly communities, integrated care, and social innovation. A structured survey conducted between January and February 2025 among 458 providers assessed structural challenges, availability of digital solutions, and feasibility of technological integration. The survey was carried out within the CENTINOSS project and is positioned in relation to broader research on social innovations for integrated care of community-dwelling older adults and on socio-cultural and organizational aspects of knowledge and technology transfer. Descriptive analysis identified patterns of acceptance, implementation gaps, and cross-national differences. The findings show that both countries operate under similar structural constraints, particularly workforce shortages and limited financial resources, which influence the pace of digital transformation. Low-complexity communication and safety technologies, such as emergency alarms, video communication, and medication reminders, are widely perceived as feasible, whereas advanced solutions such as robotics and virtual reality remain marginal and cautiously evaluated. The study contributes to a socio-technical understanding of digital transformation in elderly care and advances a multilevel implementation perspective linking European governance, national policies, organizational capacity, and professional competencies.

Keywords: Learning Systems, Intelligent and AI Based Control, Fuzzy and Neural Systems
Abstract: 对象检测器通常性能较差 当面对域变换时的衰落 源域（收集的数据）和目标域 （真实情况 应用数据）。这是因为视觉效果显著 跨域图像的差异，例如 物体的缩放、纹理和内容风格的变化。前往 提升跨域检测性能，本文 提出积分两个模块：AssemForformer（一个 基于汇编的卷积视觉变换器）和SEAM （分离与增强注意力模块）并入RT-DETR 探测器。AssemFormer结合了局部特征提取 卷积神经网络的能力 变换金刚的全局上下文建模能力，寻址 传统卷积神经的局限性 网络捕捉长距离依赖关系和局部 细节。SEAM 提升无遮挡环境的特征响应 在补偿遮挡信息丢失的同时，区域 从而增强了 遮挡的物体。它还解

Keywords: Learning Systems, Multi-agent Systems
Abstract: Distributed Speculative Decoding (DSD) has emerged as a promising paradigm to reduce the serving costs of Large Language Models (LLMs) by offloading the drafting process to edge devices. However, the communication overhead required to synchronize probability distributions between the cloud server and the edge client creates a significant bottleneck, particularly in bandwidth-limited environments. Existing solutions often resort to lossy compression methods, which degrade the probabilistic integrity required for complex reasoning tasks. In this paper, we propose a novel Two-Stage Rejection Sampling framework. Our approach exploits the heavy-tailed nature of residual probability distributions by partitioning the vocabulary into a dominant "Head" and a sparse "Tail." By prioritizing the transmission of the Head and employing a hierarchical rejection sampling scheme, we drastically reduce data transmission while mathematically guaranteeing exact sampling from the target distribution. Experimental results show that our method effectively alleviates bandwidth pressure and accelerates inference without compromising the model's reasoning fidelity.

Keywords: Networked Control, Linear Systems, Motion Control
Abstract: Addressing the problem of spacecraft attitude formation control, this paper proposes an innovative approach: unlike traditional methods, short-duration jet-driven signals are approximated as impulsive control signals to characterize the “instantaneous” changes in the spacecraft’s angular velocity. Furthermore, instead of pursuing equivalence with continuous control signals, the evaluation of impulsive control effectiveness is directly based on closed-loop control performance, aiming to design an impulsive controller that ensures the closed-loop system performance approximates that of a given nominal continuous reference system. This method provides a new theoretical framework for significantly reducing spacecraft fuel consumption and actuator wear while maintaining control accuracy. Finally, numerical examples are presented to demonstrate the effectiveness of the proposed control algorithm.

Keywords: Adaptive Control, Control Applications, Robotics
Abstract: Safe and adaptive assistance remains a key challenge for lower-limb rehabilitation exoskeletons, because the required human--exoskeleton interaction stiffness should vary continuously with both the instantaneous tracking deviation and the user’s recovery state. Conventional fixed-impedance or discretely switched controllers either over-constrain natural gait or provide insufficient corrective guidance when deviations grow. To address this problem, this paper proposes a state-scheduled regional variable impedance control (RVIC) scheme. RVIC introduces a sigmoid-based virtual safety corridor in the error space to regulate the joint stiffness smoothly: low stiffness is maintained inside the corridor to preserve transparency, while stiffness increases continuously outside the corridor to provide corrective assistance. In addition, a recovery score sin[0,1] is incorporated as a scheduling variable to continuously tune both the overall assistance intensity and the corridor width, enabling a principled assistance--autonomy trade-off across rehabilitation stages. Simulations and treadmill experiments on a wearable lower-limb exoskeleton with five healthy subjects validate that RVIC achieves stable human--exoskeleton interaction and smooth stiffness transitions.

Keywords: Adaptive Control, Control of Smart Power Delivery Systems, Energy Efficiency
Abstract: Abstract—The application of Proton Exchange Membrane Fuel Cell (PEMFC) Combined Heat and Power (CHP) systems in cold regions presents significant energy-saving potential, yet their dynamic thermal management remains a critical challenge. The cooling loop of the CHP system is characterized by strong nonlinearity, large thermal inertia, and high susceptibility to severe disturbances from internal electrical loads and external extreme-cold heat grids. Traditional controllers often suffer from integral windup and sluggish responses under such conditions, leading to severe temperature overshoots, "cold shock" risks, and excessive parasitic power consumption. To address these issues, this paper proposes a novel Adaptive Smith Predictor-based Active Disturbance Rejection Control (Adaptive SP-ADRC) strategy. And Forgetting Factor Recursive Least Squares (FFRLS) algorithm is integrated to online identify the time-varying high-frequency gain, ensuring the controller adapts to physical parameter drifts. Concurrently, a Smith Predictor is employed to compensate for the large thermal lag, enabling the linear extended state observer (LESO) to effectively estimate and reject multi-source disturbances. A high-fidelity thermodynamic model is established and tested under extreme dynamic scenarios. Simulation results demonstrate that, compared to a baseline Fuzzy-PID controller, the proposed Adaptive SP-ADRC reduces the maximum temperature overshoot by 81.5% during electrical load steps and effectively prevents "cold shock" during sudden external cold impacts. Index Terms— Proton exchange membrane fuel cell (PEMFC), Combined heat and power (CHP), Thermal management, Active disturbance rejection control (ADRC), Online parameter identification, Large thermal inertia.

Keywords: Adaptive Control, Fuzzy and Neural Systems, Multi-agent Systems
Abstract: This paper addresses the cooperative target enclosing problem for underactuated autonomous surface vehicles (ASVs) under jointly connected directed graphs. To tackle the challenges of limited communication and time-varying, intermittently disconnected topologies in practical maritime environments, a fully distributed event-triggered control framework is proposed, which consists of a coordination layer and a control layer. In the coordination layer, an event-triggered adaptive fully distributed observer is designed, enabling each ASV to achieve asymptotic estimation of both the system matrix and the state of the moving target under intermittent communication while strictly excluding Zeno behavior. Based on this, the control layer generates distributed guidance commands in the Earth-fixed frame according to the desired enclosing distance and relative angle. It then employs fuzzy logic systems to approximate the unknown vessel dynamics and designs a robust tracking controller to ensure the accurate execution of the guidance commands. Simulation results demonstrate that the proposed method ensures the enclosing errors are uniformly ultimately bounded under communication constraints and topology switching, while significantly reducing the communication frequency.

Keywords: Adaptive Control, Intelligent and AI Based Control, Modeling and Control of Complex Systems
Abstract: Аннотация. В этой статье описывается разработка и тестирование интеллектуальной системы управления, предназначенной для оптимизации работы электростатических осадочных средств в реальном времени. Основная цель исследования — повысить эффективность очистки дымовых газов с помощью интеллектуальных алгоритмов на основе ис

Keywords: Adaptive Control, Multi-agent Systems, Linear Systems
Abstract: The existing bearing-based formation control has mainly focused on multiple integrator systems. The main reason is that the existing methods use Routh's criterion to determine various control gains, which is only effective for low order integrator systems. In this paper, we further study the bearing-based formation control for a large class of linear multi-agent systems. By leveraging the solution of a particular Riccati equation, we manage to find a distributed bearing-based control law to achieve the bearing-based formation. Moreover, we introduce an adaptive bearing-based control technique to deal with the disturbance rejection problem for bounded disturbances with unknown bounds. Instead of resorting to Routh's criterion, we apply rigorous Lyapunov analysis to guarantee the closed-loop stability. Finally, the effectiveness of the proposed approach is verified through the formation control of a group of quadrotors.

Keywords: Adaptive Control, Nonlinear Systems and Control, Estimation and Identification
Abstract: This paper addresses the flight control problem of high performance aircraft subject to aerodynamic parameter uncertainties and unknown control surface failures. First, the influence of control surface failures on flight performance is analyzed, and a nonlinear aircraft model incorporating actuator faults is established. Then, a parameterized model suitable for online identification is constructed, and an adaptive aerodynamic parameter estimation algorithm based on variable-rate forgetting recursive least-squares is proposed to achieve real-time parameter estimation. On this basis, a hierarchical fault-tolerant control scheme based on adaptive dynamic inversion is developed, which integrates outer loop position and trajectory control with inner loop attitude and angular rate control, enabling real-time dynamic compensation under unknown actuator faults and command tracking of flight altitude, lateral displacement and sideslip angle. Finally, numerical simulations based on an F-16 aircraft model validate the effectiveness of the proposed method. The results show that the designed parameter estimation algorithm can effectively estimate the time-varying aerodynamic parameters, and the fault-tolerant control scheme is able to maintain satisfactory tracking performance and system stability in the presence of uncertain elevator failure.

Keywords: Automated Guided Vehicles, Nonlinear Systems and Control, Motion Control
Abstract: This paper investigates the impact time control guidance problem under field-of-view (FOV) constraints. Unlike existing results, the proposed method avoids control singularity, which enhances the reliability of the guidance system. First, a guidance model is formulated and the corresponding guidance objectives are explicitly defined. Then, a nominal guidance law is developed using an inverse-dynamics-based design method to enable accurate time-to-go prediction. Subsequently, by incorporating impact-time error feedback, a nonsingular impact-time guidance law with FOV constraints is proposed, and the stability of the impact-time error is rigorously proven. Finally, numerical simulations and equivalent physical experiments are conducted to comprehensively validate the effectiveness of the proposed method.

Keywords: Sensor/Data Fusion, Estimation and Identification, Man-machine Interactions
Abstract: Latency in the sensing–estimation pipeline can make exoskeleton assistance arrive late. We study an offline time-advancing estimator that fuses sEMG and IMU signals to predict future gait phase, bilateral hip angles, and walking speed at t+δ. The model uses a lightweight dual-stream architecture with a CNN sEMG encoder, a GRU IMU encoder, and a channel-wise gating module. Evaluation on a synchronized sEMG–IMU–MoCap dataset from eight participants under six treadmill conditions (48 trials) showed that, at δ = 100 ms, the fusion model achieved NRMSE 0.082 ± 0.017 for phase, 0.060 ± 0.013 for hip angle, and 0.183 ± 0.027 for speed, with correlations of 0.970±0.014, 0.984±0.010, and 0.867±0.037. Fusion also degraded more gracefully than unimodal baselines as the horizon increased to 250 ms, supporting its use for offline future-state estimation

Keywords: Multi-agent Systems, Networked Control, Real-time Systems
Abstract: This paper presents algorithms and preliminary results for a specialized Cartographer role within the heterogeneous Khan–Syula drone swarm currently under development at the Naval Postgraduate School. The envisioned swarm comprises diverse unmanned aerial vehicles equipped with different sensors, capabilities, and mission roles, collaborating to execute complex, multi‑objective disaster‑response or defense operations. Swarm‑level efficiency is enhanced by assigning tasks such as reconnaissance, jamming, and strike to dedicated drone subsets. The Cartographer’s function is to reconstruct the geometry of the operational communication‑service region using limited flight trajectories and sparse packet‑delivery measurements. Unlike dense signal‑quality mapping, the Cartographer focuses on estimating the service‑area boundary, where trajectory‑planning and relay‑placement decisions are most sensitive. To accelerate boundary convergence, an active sampling strategy is introduced that selects measurement points along the evolving iso‑band, concentrating new observations near the current estimate. To validate the complete closed‑loop system, a hardware‑in‑the‑loop test bench was constructed, integrating a motion‑capture system with a position‑dependent virtual jammer. This jammer replicates communication losses and delays without radio‑frequency transmission, enabling controlled and repeatable comparisons among different reconstruction algorithms. Gaussian Process (GP) and Neural Field (NF)–based reconstructions are evaluated using held‑out flight runs and a dense emulator reference. Results indicate that NF yields a more stable estimate of the service‑area geometry and reduces false‑infeasible expansions near the boundary compared to GP. These improvements support the use of communication‑zone maps as feasibility constraints for planning under degrad

Keywords: Nonlinear Systems and Control, Multi-agent Systems
Abstract: This paper studies the problem of cooperative safety for multiple mobile agents moving and performing distinct tasks in a shared environment. Under information-exchange constraints, three key technical challenges must be addressed simultaneously: the distributed implementation of safety-critical controllers, guaranteed convergence rates of the optimization algorithm under communication constraints, and the preservation of safety in the presence of algorithm-induced errors. We propose a novel design of cooperative safety-critical controllers based on a distributed implementation of a dual gradient algorithm. Convergence analysis of the optimization algorithm and safety analysis of the resulting closed-loop system are developed using tools from robust nonlinear control. Sufficient conditions on the controller parameters that guarantee safety of the multi-agent system are derived, revealing a trade-off between the convergence rate of the optimization algorithm and the satisfaction of safety constraints. The effectiveness of the proposed approach is demonstrated through numerical simulation and experimental results.

Keywords: Estimation and Identification, Robotics, Sensor/Data Fusion
Abstract: Real-time and precise contact wrench estimation is essential for autonomous multirotors to enhance performance in tasks such as aerial manipulation and payload transportation. Most existing research neglects the aerodynamic interference during flights, resulting in intrinsic discrepancies between the estimated external wrench and the actual contact wrench. To address this issue, we introduce a contact wrench estimation method based on an optimal filter framework. Our approach explicitly models linear aerodynamic effects, thereby ensuring the decoupling of airflow disturbances from contact wrenches. We rigorously verify the observability of the system through Lie derivatives and present a system identification algorithm for aerodynamic parameters. Comprehensive simulations and real-world experiments demonstrate that the proposed method effectively mitigates aerodynamic interference, providing fast and accurate contact wrench estimation for multirotors.

Keywords: Robotics, Factory Modeling and Automation, Motion Control
Abstract: Robotic pick-and-place of CFRP prepregs deposits fiber angular error into the cured laminate whenever the gripper heading at pick-up deviates from the ply design axis—yet no existing automated layup system measures or corrects this deviation before contact occurs. A closed-loop, training-free architecture is presented in which wrist joint J6 is rotated to the measured fiber direction before the gripper touches the material. Fiber orientation is recovered from overhead RGB imagery via 2-D Fast Fourier Transform (FFT) analysis of tow spatial periodicity; a Peak-to-Sidelobe Ratio (PSR) gate withholds commands when spectral contrast is insufficient; and a C²-continuous quintic J6 trajectory drives the Bernoulli gripper to the target angle without perturbing the Cartesian pick point. The same pipeline accommodates unidirectional (UD) plies through single-peak spectral analysis, with peak count serving as an automatic ply-type discriminator requiring no separate classification. Closed-loop evaluation in a ROS2–Gazebo digital twin over 70°–112° at 3° increments yields MAE = 0.40° and RMSE = 0.82°; all orientations fall within the ±2° aerospace structural tolerance at 30 Hz throughput on a commodity CPU.

Keywords: Robotics, Intelligent and AI Based Control, Control Applications
Abstract: Reliable robotic manipulation in unstructured and dynamic environments remains challenging due to perception uncertainty, modelling errors, and interaction-induced disturbances. While recent vision-language models (VLMs) enable flexible instruction understanding and task decomposition, high-level semantic reasoning alone does not guarantee stable physical execution under dynamic conditions. This paper proposes a hybrid manipulation framework that integrates VLM-based task decomposition with closed-loop position-based visual servoing (PBVS) for robust task execution. High-level instructions are translated into structured manipulation primitives, which are executed through two unified control interfaces: single_arm_pbvs and dual_arm_pbvs. These interfaces regulate Cartesian pose errors using real-time visual feedback, enabling continuous error correction and stable convergence during manipulation. The proposed framework is validated on a dual-arm robotic platform in representative manipulation tasks involving dynamic disturbances and interaction variability. Experimental results demonstrate stable and consistent task execution under perception uncertainty and environmental changes.

Keywords: Robotics, Modeling and Control of Complex Systems
Abstract: The multi-fingered robot hand is a generalizable device that acts as the end effector of robot manipulators or humanoid robots to perform various dexterous tasks, e.g., grasping in clutter and in-hand manipulation. The multi-fingered robot hand is commonly equipped with multiple degrees of freedom (DoFs). While such high DoFs lay the foundation for highly skillful tasks, they also open up challenges for the modeling of multi-fingered hands, which is very important as it describes the mapping from the motion of finger joints to the task where the robot performs. This paper proposes a new modeling method for multi-fingered robot hands, where analytical techniques are employed to establish the kinematic and dynamic models first, and then data-driven networks are constructed to compensate for the residual errors to further improve the precision. A series of simulation studies and experiments are carried out to illustrate the effectiveness of the proposed method. The results show that the hybrid approach effectively combines the interpretability of analytical modeling with the adaptability of data-driven learning, significantly improving pose estimation accuracy and enhancing performance in dexterous manipulation tasks while maintaining physical plausibility.

Keywords: Robotics, Motion Control
Abstract: Torso-free legged systems are lightweight and mechanically simple. In single support, however, stance-driven motion induces reaction torques on the other leg, so the shape dynamics are strongly coupled and cannot be arbitrarily shaped by direct actuation. This study revisits a planar two-rod minimal robot as a testbed for underactuated legged-rolling and develops an acceleration-level orbit-tracking framework. By writing the constrained dynamics in a projection form and deriving an explicit input–output acceleration map for a three-dimensional shape output, we design a computed-acceleration tracking law that commands the desired output accelerations and incorporates PD feedback. Since touchdown induces impacts and leg relabeling, step-to-step variability is inherent to the hybrid dynamics. Accordingly, we examine how orbit design, i.e., the time scale and terminal geometry, together with feedback gains, shape the emergent gait and its asymptotic periodicity. The proposed formulation provides a compact and interpretable platform for studying gait formation in a torso-free, highly underactuated morphology.

Keywords: Robotics, Motion Control, Automated Guided Vehicles
Abstract: Trajectory tracking of mobile robots under un known time-varying disturbances and nonstationary noise is challenging. This paper presents a disturbance-aware control framework combining a nonlinear PID controller with feed forward disturbance compensation. Linear and angular disturbances are modeled as augmented states and estimated online using EKF and Adaptive EKF (AEKF). While EKF assumes fixed noise covariances, the AEKF updates them adaptively using innovation statistics to maintain estimator consistency under abrupt noise changes. A circular reference trajectory ensures persistent excitation, and sudden noise variations are introduced to assess robustness. Simulation results show that EKF improves tracking under stationary noise but degrades under nonstationary conditions. In contrast, the AEKF maintains bounded estimation error, accurate disturbance reconstruction, and superior tracking performance, highlighting the importance of adaptive covariance tuning for robust nonlinear mobile robot control.

Keywords: Robotics, Motion Control, Control Applications
Abstract: This paper presents the design, modeling, and experimental validation of an underwater vehicle equipped with a reaction wheel system for high-precision attitude control. Unlike underwater platforms that rely on hydrodynamic forces generated by rudders or differential thrust, the proposed system achieves attitude control through internal momentum exchange, thereby eliminating the need for external flow interaction. A complete dynamic model of the vehicle is developed, incorporating both the rigid-body dynamics and the coupled dynamics of the reaction wheel assembly. Experimental evaluations are conducted in a water tank environment, where the proposed reaction wheel-based system is compared against a thruster-based yaw control scheme. The results demonstrate that the proposed approach achieves significantly improved tracking accuracy and smoother transient behavior, highlighting its potential for precise and low-disturbance maneuvering in underwater applications.

Keywords: Robotics, Motion Control, Modeling and Control of Complex Systems
Abstract: Amphibious robots have overcome the operational limitations of single-medium environments, demonstrating significant application value in tasks such as marine monitoring and cross-domain intervention. Addressing challenges like high system complexity and limited reliability caused by drive redundancy in small amphibious platforms, this paper proposes an amphibious robot design based on “variable propeller-leg” technology and its control architecture. This configuration achieves morphological reuse of actuators, significantly reducing drive components while balancing multi-degree-of-freedom underwater maneuverability with complex terrain traversal capability on land. To enable efficient cross-medium motion control, an integrated dynamics model covering both aquatic and terrestrial environments is established, alongside a multi-mode model predictive control (MPC) framework. This framework effectively handles the highly coupled nonlinear dynamics of the robot across different media, enabling real-time optimization and coordination of control inputs for underwater and terrestrial motion while strictly adhering to physical constraints. Finally, comprehensive testing in the ROS2-Gazebo high-fidelity simulation environment validated the proposed robot's amphibious cross-domain capabilities and demonstrated the effectiveness and superior performance of the MPC control scheme.

Keywords: Fuzzy and Neural Systems, Discrete Event Systems, Real-time Systems
Abstract: The risk state of liquefied natural gas (LNG) tanker road transportation exhibits significant dynamic variability. Under continuous monitoring conditions, achieving stable and consistent risk assessment remains a critical challenge. To address this issue, this paper establishes a comprehensive risk indicator system and proposes a dynamic fuzzy evaluation model driven by a sliding-window-enabled adaptive fusion mechanism. Specifically, this mechanism dynamically reconciles expert prior knowledge with streaming data characteristics to enable adaptive weight adjustment. By updating indicator membership degrees based on monitoring data, multi-source risk indicators are mapped to the probability distribution of the vehicle’s overall risk level, thereby realizing continuous dynamic risk assessment. Experimental results show that, under an optimal 60-second sliding window, the proposed model effectively reflects the dynamic evolution of the risk estimation process. Under 20 dB noise interference, the posterior probability vector yields an average root mean square error (RMSE) of 0.0269 and a maximum RMSE of 0.1998, demonstrating strong robustness against sensor disturbances and the ability to stably and adaptively track changes in estimated risk states.

Keywords: Fuzzy and Neural Systems, Micro and Nano Systems, Estimation and Identification
Abstract: An improved quantum neural network based on a parameterized quantum circuit for 0 or 1 classification of images is designed in this paper. A classical handwritten digital image of size 28×28 pixels is compressed into a 4×4 pixel image, whose 16 classical images represent the probability amplitudes of a 4-qubit state input in the quantum state preparation circuit. The output and corresponding 0 or 1 labels are used as training data for supervised learning of a parameterized quantum recognition circuit composed of 4 qubits. The complete design process is studied in detail, including the input of classical data, quantum state preparation, the output recognition of images by the parameterized quantum circuit, and the projection measurement of the designated readout qubit. In the training of the network parameters, the goal of minimizing the loss function is to achieve optimal quantum circuit parameters. In the performance comparison experiments, handwritten digital images from the MNIST dataset are used to conduct recognition tests on two specific digits, and the test accuracy is 98.88%. The recognition accuracy rate increases by 8.13% compared with the recognition performance before the improvement. The quantum image recognition network designed in this paper can be extended to multiclass classification applications.

Keywords: Modeling and Control of Complex Systems, Robotics, Adaptive Control
Abstract: 机器人辅助喂养系统对于提升上肢障碍者的饮食独立性至关重要。然而，现有的喂食机器人在铲取时稳定性差且近口接触时安全性低，难以广泛应用。为应对这些挑战，我们提出了一个稳健的层级控制和统一评估框架，通过构建一个结合有限状态机（FSM）引用生成与PPO残差控制的分层架构，并建立利用仿真特权信息的列车-评估分割协议，实现“食物舀取到送达”过程的统一评估框架。模拟结果表明，所提方法在两种颗粒负载（分别为300和800个颗粒）下，任务成功率分别为0.380和0.420。这比传统的有限状态机策略Park-FSM（0.120/0.140）大约提升了三倍，并且明显优于Traj-Improve（0.180/0.220）。在两种负载Ç

Keywords: Robust and H infinity Control, Modeling and Control of Complex Systems, Linear Systems
Abstract: Small unmanned aerial vehicles (UAVs) flying close to the ground and structures face spatially heterogeneous disturbances, such as shear layers, wake recirculation, and varying ground effects. These disturbances can amplify closed-loop position errors and control effort. Although wind and turbulence can be measured using specialized sensors or estimated through modeling techniques, there is currently no method to quantify closed-loop control robustness in space using standard flight logs alone. This paper proposes the Airspace Safety Index (ASI), a bounded scalar field derived from PX4 flight logs that maps three-dimensional space to values between 0 and 1. In this work, safety is defined operationally as a low likelihood of large destabilizing motion within a given airspace for a specific UAV and controller configuration. ASI combines disturbance-sensitive dispersion features from flight logs, including attitude jitter, angular-rate jitter, horizontal acceleration, and motor-command variability. These features are combined using robust percentile normalization and convex weighting to produce a spatial representation of reduced closed-loop control robustness. The method is validated using proxies for drift radius and closed-loop motion, and a reproducible artifact bundle and pipeline are provided for generating tables and spatial maps from standard flight logs. ASI is further evaluated through stress tests, including dose-response gust injection in simulation, controller intervention, and spatial repeatability tests, to address concerns related to estimator coupling and overfitting.

Keywords: Optimal Control, Robotics, Nonlinear Systems and Control
Abstract: This article presents a Hamilton–Jacobi (HJ) reachability framework for a two–satellite collision avoidance problem operating in the same circular orbit, where relative motion is modeled in the radial–tangential–normal (RTN) frame using planar Hill–Clohessy–Wiltshire (HCW) dynamics. We define the target state space as unsafe relative configurations in the orbit plane corresponding to minimum separation requirements consistent with Federal Communications Commission (FCC) orbital standards. The interaction between spacecraft is formulated as a zero–sum differential game, where Player 1 is the controlled satellite and Player 2 is modeled as a bounded adversarial disturbance with unknown intent. We present the HJ formulation and compute backward reachable sets that characterize relative states from which collision cannot be avoided under worst-case disturbances, while states outside this set admit provably collision-free trajectories. These reachable sets are integrated with supervisory hybrid control logic to determine when evasive maneuvers must be initiated, enabling mathematically grounded safety guarantees for scalability.

Keywords: Robust and H infinity Control, Nonlinear Systems and Control, Optimal Control
Abstract: The proportional-integral-derivative (PID) controller and its variants are widely used in control engineering, but they often rely on linearization around equilibrium points and empirical parameter tuning, making them ineffective for multi-input-multi-output (MIMO) systems with strong coupling, intense external disturbances, and high nonlinearity. Moreover, existing methods rarely explore the intrinsic stabilization mechanism of PID controllers for disturbed nonlinear systems from the perspective of modern robust control theories such as dissipativity and mathcal{L}_2-gain. To address this gap, this study focuses on gamma-dissipativity (partially equivalent to mathcal{L}_2-gain) and investigates the optimal parameter tuning of MIMO-PI controllers for general disturbed nonlinear MIMO systems. First, by integrating dissipativity theory with the Hamilton-Jacobi-Isaacs (HJI) inequality, sufficient conditions for the MIMO-PI-controlled system to achieve gamma-dissipativity are established, and the degree of gamma-dissipativity in a local region containing the origin is quantified. Second, an optimal parameter tuning strategy is proposed, which reformulates the gamma-dissipativity optimization problem into a class of standard eigenvalue problems (EVPs) and further converts it into linear matrix inequality (LMI) formulations for efficient online computation. Comprehensive simulation experiments validate the effectiveness and optimality of the proposed approach. This work provides a theoretical basis for the robust stabilization of general disturbed nonlinear MIMO systems and enriches the parameter tuning methods of PID controllers from the perspective of dissipativity.

Keywords: Control Applications, Motion Control, Robotics
Abstract: Existing quadcopter-based search and rescue (SAR) operations have not yet sufficiently integrated onboard active perception with real-time motion control, which significantly limits overall search efficiency. This paper introduces an active radio frequency (RF) source-seeking framework utilizing a quadcopter quadcopter as a temporary long-term evolution (LTE) base station. The quadcopter follows a controlled helical trajectory around a moving orbit center, collecting Power Headroom (PHR), Signal-to-Noise Ratio (SNR), and Modulation and Coding Scheme (MCS) measurements. These are reconstructed into local scalar signal strength fields using angular-domain zero-order hold and closed-path integrals to estimate 3D gradients. Fidelity-weighted fusion of the gradients generates a steering vector for the orbit center, enabling gradient-ascent navigation toward the source. A geometric controller defined on the special Euclidean group SE(3) directly computes thrust and moment inputs to achieve stable helical motion, ensuring agile and robust tracking suitable for time-critical SAR operations. Theoretical analysis proves exponential convergence to the desired helical path segments under piecewise-constant steering. Simulations based on real-world RF field data validate the approach, demonstrating rapid error convergence, stable orbiting, and effective search to the signal source without predefined trajectories. The proposed system addresses limitations of existing SAR methods by exploiting RF signals for occlusion-resistant localization, offering an autonomous, and efficient solution that increases survival probabilities in complex terrains.

Keywords: Factory Modeling and Automation, Real-time Systems, Multi-agent Systems
Abstract: In dynamic environments, multi-UAV systems encounter significant challenges in executing coverage missions due to uncertainties such as unknown obstacles, sudden task requirement changes, and equipment failures. Traditional offline planning methods struggle to adapt to dynamically changing environments. Therefore, to address the multi-region coverage problem in multi-UAV systems under scenarios with dynamically changing task regions, we propose a novel D2CP (Discretization, Conversion, Partitioning, Conflict Resolution, Path Planning) algorithm. This algorithm incorporates multiple dynamic factors and is capable of effectively adapting to real-time changes in UAV status and task requirements, demonstrating strong adaptability and practicality. The D2CP algorithm employs a divide-and-conquer strategy, dividing the process into five stages to efficiently handle task replanning. Experimental results using DJI Phantom 4 Pro and Matrice 300 RTK drones demonstrate significant reductions in solving time and task completion time, while achieving balanced workload distribution among UAVs. Our approach shows robust performance across various dynamic scenarios, confirming its effectiveness in complex environments.

Keywords: Multi-agent Systems, Networked Control
Abstract: In this article, the generalized distributed average tracking (DAT) problem for multi-agent systems is investigated over diverse detection networks. Traditional DAT algorithms assume that each agent can detect a reference signal, which is restrictive in practical scenarios where agents possess diverse detection capabilities. First, an embedded generalized DAT algorithm is proposed to estimate the average of the detectable reference signals. By distributively estimating network characteristic information and embedding it into the average reference signal estimator, the generalized DAT problem can be effectively addressed even when some agents are unable to detect the reference signal. Furthermore, to handle more diverse detection tasks, a class of generalized DAT algorithms over diverse detection networks is developed, which allows each agent to detect an arbitrary number of reference signals. Compared with existing DAT algorithms, this article is the first to consider the generalized DAT problem applicable to diverse detection scenarios, including one-to-one, one-to-many, many-to-one, and many-to-none cases, thereby relaxing the detection capability limitations imposed by traditional DAT algorithms. Finally, simulation examples are provided to verify the effectiveness of the proposed algorithms.

Keywords: Multi-agent Systems, Nonlinear Systems and Control
Abstract: This paper addresses the asymptotic stability and the estimation of the domain of attraction (DOA) for the directed heterogeneous complex network. Assuming that the network is strongly connected, we find that if there exists an individual system whose linearized system is asymptotically stable with respect to its equilibrium point, then there must exist appropriate coupling strengths such that the entire network is asymptotically stable with respect to its equilibrium point. Due to excessively high computational complexity, many algorithms for estimating the DOA by iteratively computing Lyapunov functions fail to perform effectively on complex networks. To avoid this, we choose the Lyapunov function of the linearized system of the network for estimating the DOA and then use a binary search-based approach to enlarge this estimation. Moreover, for network with polynomial vector field, we convert the computation of the estimation of the DOA into a typical problem of solving sum of squares (SOS) programming. Especially, by decreasing the number of parameter variables and the degree of the polynomials in the SOS programming, the computational complexity is significantly reduced such that the estimation of large scale network also can be solved by the existing semi-definite programming software. Finally, three examples are used to illustrate the validity and effectiveness of our approach.

Keywords: Multi-agent Systems, Robotics
Abstract: With the increasing importance of autonomous aerial vehicles (AAVs), the AAV coverage path planning problem has attracted increasing attention. This paper focuses on multi-AAV multi-regional coverage path planning with a maximum range constraint. To address this problem, this paper proposes a new modeling approach by dividing task regions into strips and reformulating the problem as a clustered multi-depot multiple traveling salesman problem with a maximum range constraint. A two-stage hybrid algorithm based on adaptive large neighborhood search is developed to solve the problem. In the first stage, an initial solution is generated using a marginal insertion cost-based constructive heuristic. In the second stage, an ALNS framework combined with local search iteratively destroys and repairs solutions to escape local optima. Considering the cluster constraint, cluster (strip)-specific operators are designed by extending vertex operations to clusters and adjusting the starting and ending vertices of each cluster through random changes or swaps. Experimental results show that the proposed algorithm produces better solutions than the comparison algorithms for most test instances.

Keywords: Multi-agent Systems, Sensor/Data Fusion, Estimation and Identification
Abstract: 动态无人机（UAV）集群 面临诸如通信带宽有限等挑战， 协作感测时的剧烈机动。要处理 本文提出了这些问题的层级聚会 分布式合作本地化框架。该 定位过程分为两个层级，基于 相对测量。在集群内部层面，成员 节点利用高频到达角（AOA）数据 通过分布式加权共识获取局部估计值。 在星系团间层面，星系团会周期性地出现首位 引入到达时间差（TDOA）约束 进行全局几何细化。此外，还有 扩展卡尔曼滤波器（EKF）被集成以熔合这些 用目标运动模型进行空间估计，以实现稳定 追踪。模拟结果表明，所提出的 算法显著降低了通信开销 同时实现了与 集中式方法。

Keywords: Multi-agent Systems, Networked Control
Abstract: This paper studies leader-follower formation tracking problem for hybird multiple nonholonomic robots with visibility constraints. With onboard vision sensors on follower robots and inter-robot communication, a new distributed controller for leader-follower formation is proposed based on dynamic surface control method. Our presented controller can make tracking errors of relative distances and bearing angles between each leader-follower pair in the robot team arbitrarily small by adjusting parameters. In addition, the visibility sensor constraints and connectivity of communication graph are preserved all the time. Unlike most of papers which can merely cope with the condition where one leader tracks only one leader, our methods can be suitable to more general network structure with multiple leaders as well. Furthermore, the proposed method is adaptive to different kinds of robots including first-order and second-order multi-robot systems. Also, it can be robust to actuator faults and abrupt disturbance in second-order robots. Lyapunov stability analysis and several numerical simualtion are presented to verify the correctness and effectiveness of our proposed controller.

Keywords: Automated Guided Vehicles, Learning-based Control, Optimal Control
Abstract: Trajectory tracking at handling limits poses a critical challenge for autonomous driving systems, where parameter uncertainties and highly nonlinear tire dynamics necessitate adaptive control. However, standard adaptive implementations often exhibit instability. For instance, while Gaussian Process Model Predictive Control (GP-MPC) can theoretically correct model mismatches, standard implementations suffer from overfitting-induced control chattering driven by unconstrained likelihood maximization on noisy data. In this paper, we present a Bandwidth-regularized Sparse GP-MPC scheme to address this issue by combining a Subset of Data (SoD) Sparse GP approximation with a physically-consistent hyperparameter clamping strategy. Specifically, bounding the kernel length-scale imposes a strict spectral bandwidth limit. This essentially prevents the model from fitting high-frequency noise beyond the physical limits of the actuators. Co-simulation tests confirm that our regularized approach resolves the severe control chattering seen in baseline GP-MPC. The vehicle recovers a smooth transient behavior that closely matches the nominal reference, yielding reduced overshoot, faster settling times, and better overall tracking precision.

Keywords: Control of Distributed Generation Systems, Learning-based Control, Optimal Control
Abstract: The integration of high proportions of distributed photovoltaic (PV) systems has led to voltage violations in distribution power grid. Using the remaining capacity of PV inverters for reactive power regulation is an efficient voltage control method. Existing studies typically address PV reactive power capabilities statically, failing to fully account for dynamic variations caused by solar radiation fluctuations. This paper proposes a data-driven probabilistic assessment and hierarchical reactive power optimization method. A spatiotemporal probabilistic prediction model is constructed to perform multi-quantile predictions of PV active power and node loads, and dynamically constructs reactive power support boundaries to assess PV reactive voltage support. A hierarchical optimization strategy is developed to coordinate the adjustment of various reactive power resources in the distribution power grid. Case study results using an improved IEEE 33-bus system show that the proposed method effectively mitigates voltage fluctuations and improves the efficiency of reactive power regulation.

Keywords: Learning-based Control, Control Applications, Learning Systems
Abstract: The deployment of reinforcement learning in safety-critical domains such as robotics and autonomous driving requires agents not only to maximize task rewards but also to adhere to strict safety constraints. Safe reinforcement learning addresses this challenge by formulating constrained optimization problems, typically involving auxiliary cost functions that must remain below specified limits throughout interaction. While online safe RL algorithms have demonstrated effectiveness in controlled environments, their reliance on hazardous trial-and-error exploration poses fundamental barriers to real-world deployment. Offline safe RL emerges as a compelling alternative, aiming to learn constrained policies entirely from static, pre-collected datasets without any online interaction. This report summarizes the initial phase of an MSc research project dedicated to establishing empirical foundations in offline safe RL through systematic reproduction of key baselines, and proposes a focused future direction that leverages sequence modeling to address core challenges in the field.

We first investigate online constrained policy optimization to build intuition about constraint satisfaction mechanisms. Experiments on the SafetyPointCircle1-v0 task comparing Constrained Policy Optimization (CPO) [1] against unconstrained Trust Region Policy Optimization (TRPO [2]) yield instructive results: TRPO pursued reward maximization aggressively, achieving higher cumulative rewards at the cost of severe safety violations, while CPO maintained costs near the specified limit through second-order optimization and projection onto the feasible region. This comparison solidified understanding of how explicit constraint satisfaction can be achieved through constrained optimization rather than reward shaping.

We also explore offline safe RL through reproduction of the Constrained Actor-Critic with Policy Search algorithm on the OfflineCarCircle-v0 task.

Keywords: Learning-based Control, Linear Systems, Optimal Control
Abstract: This extended abstract addresses the robust model predictive control (MPC) problem for unknown linear systems by proposing an event-triggered model learning mechanism. First, historical system state trajectories and input sequences are utilized to estimate the unknown system matrices. An event-triggered identification mechanism is formulated to balance the computational load of online identification with the closed-loop gain. Subsequently, the min-max MPC method is implemented to achieve robust control of the unknown plant based on the identified model. Through the integration of the triggering mechanism with robust MPC, the closed-loop system is mathematically guaranteed to be robustly stable. Numerical simulation validates the effectiveness of the proposed method.

Keywords: Learning-based Control, Optimal Control, Learning Systems
Abstract: 问题（OCP-LS），本文系统性地研究 哈钦森迹的多重改进策略 估计。 这些包括将单一抽样与 指数平滑，降低估计频率， 改变随机向量的分布，并使用 按区块进行哈钦森估计。 使用 ResNet-18 作为 CIFAR-10图像分类任务中的基准模型，我们对收敛进行比较分析 不同改进策略在 训练预算有限。实验结果显示， 每步仅有一个Hutchinson样本，引入适当的平滑处理和结构化估计策略 可以显著降低估计方差并实现 早期训练中更快更稳定的收敛 舞台与原始哈钦森方法的比较。 进一步 实验表明，一些改进的方法可以实现 收敛速度可与 受约束下的随机梯度下降（SGD） 计算资源。我们的发现表明 对哈钦森Ą

Keywords: Adaptive Control, Intelligent and AI Based Control, Learning-based Control
Abstract: Unmanned aerial vehicle (UAV)-based dynamic traffic monitoring requires real-time adaptive trajectory planning to track spatially distributed and temporally evolving demands in urban environments. However, existing reinforcement learning (RL)-based approaches encode heterogeneous state information through shared networks, neglecting the temporal patterns in trajectory histories and the inter-target interactions among dynamic demands, resulting in policies prone to detours and ineffective prioritization. In this paper, we identify that the monitoring state naturally decomposes into two structurally distinct components: ordered trajectory sequences that reflect motion dynamics, and unordered target sets that capture demand interactions. To this end, we propose AdaUMon, which incorporates a dual-branch trajectory module with spatial and motion encoders to capture complementary movement patterns, and a self-attention target module to model inter-demand interactions and dynamic priority. The learned representations are fused into a unified state embedding for PPO-based policy optimization. Extensive experiments across two urban scenarios show that AdaUMon reduces travel distance by up to 43% compared to conventional re-planning baselines while achieving full compliance with communication-restricted zones.

Keywords: Control Applications, Adaptive Control, Modeling and Control of Complex Systems
Abstract: This paper presents a comprehensive Digital Twin (DT) framework for hydroelectric systems that enables adaptive control of both turbine governors and excitation systems without requiring detailed manufacturer specifications. The proposed framework integrates neural network-based system identification with stabilizing adaptive control laws for the installed turbine controller and middle-branch adaptive tuning for installed voltage regulator. Using real operational data from Unit C-8 at Rocky Reach Dam (1,349 MW capacity), high-fidelity neural network models are developed to capture turbine and generator dynamics without requiring detailed manufacturer specifications. The DT enables safe controller synthesis and validation in simulation before deployment. For turbine control, the proposed method achieves 79.9%, Mean Square Error (MSE) reduction compared to the optimal controller. For voltage regulation, the adaptive excitation controller achieves approximately 42.6% MSE reduction while preserving installed protection logic. The results demonstrate that digital twin technology provides a practical pathway for modernizing hydropower control systems with minimal operational disruption.

Keywords: Control Applications, Modeling and Control of Complex Systems, Intelligent and AI Based Control
Abstract: To address the difficulty of balancing vehicle speed tracking and fuel economy for heavy-duty trucks under cruise control on undulating roads, this paper proposes an energy-saving cruise control method based on offline reinforcement learning, aiming to jointly minimize speed deviation and fuel consumption. To mitigate the overestimation of Q-values for out-of-distribution actions caused by distributional shift in offline data, Conservative Q-Learning (CQL) is introduced to enhance the reliability of value estimation, thereby improving the stability and deployability of the learned policy. Simulation results show that, compared with conventional control methods, the proposed approach can suppress speed fluctuations more rapidly under real road-grade disturbances and effectively reduce cumulative fuel consumption. In addition, hardware-in-the-loop experiments verify that the method meets high-frequency real-time control requirements and demonstrates strong potential for practical engineering applications.

Keywords: Control Applications, Motion Control, Real-time Systems
Abstract: Coordinate measuring machines (CMMs) are critical equipment in the field of high-end industrial geometric metrology. They are widely used in precision manufacturing and aerospace due to their high accuracy, robustness, and versatility. Most existing offline trajectory planning methods rely heavily on prior knowledge of the workpiece and accurate CAD models. This dependency makes them unsuitable for blind scanning tasks, where the geometric features of the measured object are unknown. To address this issue, this study proposes a 3D real-time trajectory planning and correction algorithm based on probe feedback, which breaks the traditional dependency on CAD models and prior information. It enables real-time planning and dynamic correction of the probe path, even in the absence of sufficient prior information. The theoretical derivation and design process of the algorithm are detailed. Experimental results show that the proposed algorithm demonstrates good feasibility, stability, and robustness when dealing with complex surfaces with unknown shape features. The generated trajectory effectively covers the target area, and the speed, acceleration, and probe deformation of each axis remain within the set limits, indicating engineering feasibility and potential for broader applications.

Keywords: Control Applications, Optimal Control, Estimation and Identification
Abstract: Experimental validation in structural dynamics is essential for engineering education, particularly in earthquake-prone regions like Peru. However, the high cost of commercial shake tables limits their availability in many university laboratories. This paper presents the design, modeling, and control of a low-cost (approx. 450 USD) single-axis seismic simulator. The system utilizes an ESP32 microcontroller, a DC motor, and a MATLAB-based interface to reproduce customizable vibration profiles. To ensure accurate trajectory tracking of reference seismic signals, a Linear Quadratic Regulator (LQR) combined with a full-order state observer was implemented. Experimental results demonstrate that the controller effectively manages the system’s mechanical inertia, making this low-cost prototype a highly viable and accessible pedagogical tool for structural engineering laboratories in developing regions.

Keywords: Control Applications, Robotics, Automated Guided Vehicles
Abstract: Autonomous vehicles rely on integrated perception, planning, and control systems to operate safely and efficiently. Traditional systems often depend on powerful computing hardware, but do not perform as well on smaller embedded platforms. To address this, we developed an end-to-end solution which can achieve accurate real-time decision-making using lightweight algorithms. The objective was to design a system that operates efficiently and delivers the fastest possible response on resource-constrained platforms. This paper presents a real-time, vision-based autonomous driving system developed for the 2025 American Control Conference (ACC) Self-Driving Car Student Competition using QCar 2, a small-scale autonomous vehicle platform developed by Quanser. The perception module was designed using a YOLOv8 deep learning model trained to detect stop signs, traffic lights, and cones from RGB-D and CSI camera feeds. The final model achieved an overall mAP@0.5 of 0.979, with class-specific results of 0.985 for stop signs, 0.975 for red lights, and 0.977 for cones, ensuring high detection precision and reliability. For navigation, the system employed the A* algorithm to generate optimal paths and a PID-based Pure Pursuit controller for accurate and smooth trajectory tracking. The system was validated in a competitive setting at the ACC 2025 Self-Driving Car Student Competition in Denver, USA, where it secured first place.

Keywords: Control of Distributed Generation Systems, Control of Smart Power Delivery Systems, Control Applications
Abstract: Distributed control of DC Microgrids is gaining widespread application in modern power grids. However, the sparse communication network is vulnerable to cyber-attacks. In this paper, a super-twisting sliding mode observer-based adaptive distributed attack-resilient control strategy is proposed for DC Microgrid in the presence of unbounded attacks injected into the control input channel. The super-twisting algorithm eliminates estimation errors and guarantees fast convergence of the system. Moreover, to automatically adjust the compensation strength, an adaptive feedback law is further designed. The effectiveness of the proposed strategy is verified by simulations via MATLAB/Simulink.

Keywords: Robotics, Control Applications, Motion Control
Abstract: Offshore wind power is a key contributor to the global transition toward renewable energy. However, offshore wind turbine operation and blade inspection—particularly in harsh marine environments—remain challenging due to high costs and safety risks. This study presents an aerial manipulator specifically designed for wind turbine blade inspection. The proposed system integrates three components: a bio-inspired adaptive end-effector, a force-feedback telescopic manipulator, and an anti-disturbance control framework. The end-effector, inspired by the lotus seedpod, improves surface conformability and contact stability, reducing slippage and transient contact loss under vibration. The telescopic manipulator absorbs impact loads and vibrational energy, enhancing operational safety and mechanical reliability. A hierarchical control architecture incorporating a nonlinear disturbance observer (NDOB) is implemented to ensure platform stability during physical interaction. The framework enables accurate attitude regulation and trajectory tracking of the UAV under dynamic environmental disturbances. The system is validated through simulations and physical scenario experiments.