Keywords: AI-powered robotics, Robotic learning and adaptation
Abstract: Large models based on internet-scale data can now pass the Turing Test for intelligence. In this sense, data has "solved" language and many analogously claim that data has solved speech recognition and computer vision. Will data also solve robotics and automation, allowing general-purpose humanoid robots to achieve human-level performance?

Using commonly accepted metrics for converting word and image tokens into time, the amount of internet-scale data used to train contemporary large vision language models (VLMs) is on the order of 100,000 years. I'll review 3 ways researchers are pursuing to close this gap, and a 4th approach, where data is collected as real robots operate in real commercial environments – which requires bootstrapping with AI and "good old-fashioned engineering" to create robots with real return on investment that will be adopted by industry. Such robots can create a "data flywheel" to increase performance and enable new functionality, accelerating the timeline to achieve reliable, general-purpose robots.

Keywords: Cyber security networked control, Resilient networked control systems
Abstract: This paper investigates probabilistic robustness of nonconvex–nonconcave minimax problems via the scenario approach. Specifically, under convex strategy sets for all players, inspired by recent advances in scenario optimization, we first establish a probabilistic robustness guarantee for an ε-stationary point, overcoming the dependence on the non-degeneracy assumption by proving the monotonicity of the stationary residual in the number of scenarios. Furthermore, in the presence of nonconvex strategy sets, we reveal the fundamental difficulty of obtaining a tight theoretical bound based on this recent framework. Consequently, we establish a relaxed, yet rigorously valid, probabilistic bound for a global minimax point. A numerical experiment corroborates our theoretical findings.

Keywords: Multi-agent systems
Abstract: This paper investigates a multi-agent pursuit-evasion (MPE) differential game problem subject to unknown dynamics and external disturbances, where the pursuers seek to intercept the escaping evaders. The core theoretical challenge lies in determining the optimal capturing strategy for this complex game scenario. To address this, a target-selection algorithm is first introduced for pursuers, decomposing the collective MPE differential game into multiple single-pursuer-single-evader (SPSE) sub-games. Subsequently, a zero-sum differential game framework is established to derive the associated optimal game strategies. Sufficient conditions are derived to guarantee the capturability of the associated closed-loop game system. Furthermore, a data-driven reinforcement learning (RL) algorithm is developed for the online learning of the optimal game protocol. Finally, numerical simulations are conducted to validate the effectiveness of the proposed game strategy.

Keywords: Multi-agent systems
Abstract: This paper provides a new scalable verification test to ensure that disturbances do not amplify along the interconnection of a multi-agent system composed of heterogeneous agents in discrete-time. The proposed Mesh Stability extends the concept of String Stability to networks with general topology. The developed theoretical approaches are illustrated with a simulation example of a vehicle platoon in a ring road.

Keywords: Multi-agent systems, Adaptive control of multi-agent systems, Control of networks
Abstract: In multi-robot systems, maintaining persistent communication graph connectivity is often overly restrictive, especially when robots have limited communication ranges but operate in large environments. Instead, allowing robots to temporarily disconnect and later reconnect is often more desirable for efficient task execution while still ensuring timely information sharing across the team. In this paper, we propose an adaptive prescribed-time control barrier function (adaptive PT-CBF) framework that enables robots to temporarily disconnect and re-enter the communication range within an adjustable and feasible prescribed time. Moreover, we introduce a reconnection triggering mechanism that jointly considers task execution and reconnection urgency, thereby providing a principled way to decide when reconnection should occur. Theoretical analysis justifies convergence to the satisfying reconnection within a prescribed finite time. Experimental results validate the performance of our proposed adaptive PT-CBF with improved task efficiency and satisfying reconnections.

Keywords: Multi-agent systems, Control of networks
Abstract: Limitations on the power or resources available to individual agents frequently arise in real-world games. To model such situations, this work studies a class of Linear Quadratic Gaussian Graphon Mean Field Games (LQG–GMFG) whose cost functional incorporates quadratic penalties on deviations from both an agent’s privately desired control and its local control mean field. These penalties represent two distinct motivations: individualism (acting on private preferences) and conformism (avoiding the higher resource costs incurred when acting differently from others). Separate state and control mean-field consistency conditions are imposed, and conditions for the existence of solutions are given. Using spectral decomposition, an explicit value function is obtained for the infinite-horizon, exponentially discounted stationary case, and numerical simulations reveal a trade-off between conformism and individualism.

Keywords: Multi-agent systems, Control over networks, Distributed control and estimation
Abstract: Affine formation control provides an efficient framework for global maneuvers, but it is challenged by local, non-affine deformations. Such deformations induce high internal stress within conventionally rigid interaction topologies, leading to increased control effort. Inspired by structural mechanics, this paper proposes a compliant topology design method by introducing the concept of stress-energy. Specifically, we formulate two l1-regularized semidefinite programs to obtain optimal stress matrices that exhibit omnidirectional and task-specific compliance, respectively. Comparative simulations validate the superiority of our proposed topology construction schemes in reducing control cost and enhancing deformability.

Keywords: Multi-agent systems, Control under communication constraints, Consensus
Abstract: In multi-agent systems, heterogeneous time delays exist for all agents because of the difference in communication environments. Therefore, the consensus analysis of a system considering a homogeneous time-varying delay among all agents results in conservatism. In this study, an individual-delay-reflected generalized consensus is proposed for multi-agent systems with heterogeneous time-varying delays with various bounds. To reflect heterogeneous time-varying delays, the proposed Lyapunov–Krasovskii functional is constructed by dividing the integral term into intervals containing heterogeneous delays and considering augmented vectors with delay states and integral states. Furthermore, by adding zero equality conditions, conservatism is reduced. N-dependent generalized integral inequality is used to allow the user to adjust the computational complexity. Numerical examples demonstrate a reduction in conservatism with the proposed consensus criterion.

Keywords: Multi-agent systems, Distributed control and estimation
Abstract: We study multi-agent systems composed of linear agents interconnected through state coupling and subject to external disturbances. Considering a broad class of network topologies without imposing structural restrictions, we describe the overall system dynamics using a matrix-weighted adjacency matrix. Building on conditions that guarantee a bounded L2 gain for a given network, we derive sufficient conditions under which an entire family of networks achieves a scalable L2 gain, i.e., a performance bound that remains independent of network size. These results provide a systematic framework for assessing robustness and scalability in dynamically varying multi-agent networks with MIMO agents. The results are illustrated by a numerical example.

Keywords: Multi-agent systems, Distributed control and estimation, Consensus
Abstract: Most formation control methods emphasize controller design while overlooking reference formation generation, which is crucial for collaborative performance and safety. This paper proposes a safety-aware formation generation and control framework that enables flexible multi-agent maneuvering in complex environments with dual-layer safety guarantees. First, we introduce parameter-level control barrier function (CBF) that imposes safety directly in the affine-parameter space, ensuring the generated reference formation is inherently collision-free. Then, a distributed consensus algorithm is proposed to drive all agents to consensus on common affine parameters, yielding coherent formation deformations. Finally, a standard agent-level CBF-based quadratic program is employed as a backend controller to track the safe reference trajectories. Simulations in cluttered environments validate the effectiveness of the approach.

Keywords: Multi-agent systems, Distributed control and estimation, Consensus
Abstract: This paper proposes a decentralized control framework for collision-free trajectory tracking in homogeneous multi-agent systems with actuation constraints. Building on the concept of adaptive performance functions known for single agents, the method enables each agent to autonomously regulate its transient tracking performance in response to local interactions and control saturation. The core contributions are a dynamic consensus-based reference generation mechanism and a relevance-based selection of potential collision partners using a prediction of the closest approach. A modified flexible performance law ensures that tracking performance is preserved even when avoidance or saturation temporarily dominate the control action. A Lyapunov-based analysis guarantees invariance of the performance envelope and boundedness of all closed-loop signals. Simulation results with interacting agents in a three dimensional space demonstrate collision-free motion and convergence.

Keywords: Multi-agent systems, Distributed control and estimation, Control of networks
Abstract: This paper addresses the problem of distributed stabilization for heterogeneous linear multi-agent systems (MASs). It is assumed that all agents use relative state/output information, while only a subset can utilize absolute measurements. We present a Lyapunov-based approach, proposing both state- and output-feedback protocols. Under the standard stabilizability and detectability assumptions, it is shown that the proposed protocols ensure distributed asymptotic stabilization if the directed augmented communication graph contains a spanning tree. The effectiveness of the proposed approach is demonstrated through a simulation example, which verifies the ability of the proposed control strategy to stabilize heterogeneous linear MASs under the specified conditions.

Keywords: Multi-agent systems, Distributed control and estimation, Control of networks
Abstract: This paper studies the problem of cooperative multi-target enclosing. More specifically, we propose a distributed control framework to address the case where it is necessary to split or merge the team of agents as the distance between target groups increases or decreases, respectively. We first present a multi-target enclosing control law combining an affine formation control law with distance-based control terms to adjust formations around targets. Then, a novel weight matrix design is proposed for affine formation control of regular polygons. The distributed nature of this weight design method allows agents to locally compute the weights so that they can reorganize in subgroups or merge while ensuring convergence. Stability analysis of the proposed weight design method is included, as well as a numerical simulation using the proposed enclosing control to illustrate the splitting and merging task.

Keywords: Multi-agent systems, Distributed control and estimation, Control over networks
Abstract: Based on the practical scenario where collisions in formation control may lead to agent damage, this paper investigates the integrated problem of distance-based formation control and collision avoidance for multi-agent systems governed by port-Hamiltonian dynamics. A foundational step involves constructing a signed incidence matrix, which, by design, corresponds to a directed acyclic graph and possesses the full column rank property. To overcome the prevalent issue of local minima in traditional artificial potential fields, a novel design utilizing attraction-only potentials is introduced, with collision avoidance rigorously enforced by safety barriers. This framework leads to a unified controller that concurrently manages velocity tracking, target formation acquisition, and inter-agent safety. The stability of the resulting closed-loop system is guaranteed through LaSalle's invariance principle. Numerical simulations demonstrate the validity and effectiveness of the proposed control strategy.

Keywords: Multi-agent systems, Distributed control and estimation, Control under communication constraints
Abstract: The cooperative herding of high-entropy, non-cooperative swarms is a critical yet challenging problem in multi-agent control. However, existing macroscopic theories often rely on idealized global state availability, leading to perceptual fragmentation when applied under physical sensing constraints. To bridge this gap, we propose a Hierarchical Cooperative Perception (HCP) architecture. By coupling sparse informed observers with dense local actuators, HCP reconstructs non-local potential fields to overcome sensing blind spots without global communication. We derive a macroscopic flux balance analysis grounded in non-reciprocal field theory to establish rigorous stability conditions. Validated through large-scale simulations and high-fidelity PyBullet experiments with hundreds of quadrotors, the approach achieves an 80% higher containment rate than baseline methods. Crucially, the macroscopic formulation renders control complexity invariant to population size, ensuring scalability to massive swarms beyond hardware limits.

Keywords: Multi-agent systems, Distributed optimization, Consensus
Abstract: This paper proposes a distributed control framework for cooperative object transportation by multi-agent systems. Reference forces are computed through a constrained optimization that incorporates grasping and avoiding undesired rotation. To ensure scalability, the optimization is solved using a distributed algorithm in which each agent updates its reference force through local computation and limited neighbor-to-neighbor communication. Numerical simulations demonstrate that the proposed method maintains grasping and achieves a desired reference of the object's velocity, enabling flexible and scalable cooperative transport.

Keywords: Multi-agent systems, Distributed optimization, Control under communication constraints
Abstract: This paper presents a barrier-certified multi-agent ergodic coverage framework for safe and efficient exploration over complex non-Euclidean surfaces. We address the challenge of extending surface ergodic exploration to distributed multi-agent systems (MASs), where globally coupled ergodic statistics must be estimated cooperatively while satisfying safety and communication constraints. Building on the Laplace–Beltrami (LB) eigenbasis, we formulate a distributed ergodic coverage problem on meshable surfaces that enables cooperative exploration with respect to a desired inspection density. Safety is enforced through a unified set of control barrier functions (CBFs) guaranteeing inter-agent collision avoidance, distance-based connectivity, line-of-sight (LOS) preservation, and minimum surface clearance, leading to geometry-dependent couplings. A distributed consensus mechanism enables cooperative estimation of global ergodic statistics without centralized coordination, while maintaining performance and improving scalability. The framework is validated in a simulated 3D wind turbine inspection scenario.

Keywords: Multi-agent systems, Distributed optimization, Distributed control and estimation
Abstract: This paper studied a distributed coopetition problem for multi-agent systems (MASs), where the state of each agent reflects the extent of its contributions in a task. There are two key components in the considered coopetition problem: collaborative tasks and competitive constraints. The former necessitates a cumulative (weighted) contribution from all agents to achieve a desired outcome, while the latter comes from the competition among agents: no single agent exerts significantly more effort than the others (considering the respective weights). First, the proposed coopetition problem is transformed into an equivalent constrained optimization problem. then, a distributed algorithm for solving the coopetition problem is provided from the Karush-Kuhn-Tucker (KKT) conditions of the optimization problem. Subsequently, it is proved that the algorithm can ensure the states of agents to converge to one of its equilibria, which are the necessary and sufficient condition to the coopetition problem. Finally, an example is simulated to illustrate the effectiveness of the theoretical results.

Keywords: Multi-agent systems, Distributed optimization, Distributed control and estimation
Abstract: This paper addresses multi-robot pursuit failures caused by evader clustering, which typically leads to formation overlap and trajectory conflicts. We propose a Dynamic Adaptive Hunting (DAH) framework that replaces static assignments with a real-time dynamic clustering mechanism based on evader spatial distribution. To enhance efficiency, an intra-cluster optimization strategy refines target assignments to suppress trajectory crossings and mitigate the long-tail effect, thereby accelerating overall convergence. At the execution layer, an Artificial Potential Field (APF) controller provides goal-directed guidance with effective collision avoidance. Simulations across varying swarm scales confirm that DAH significantly reduces capture time and travel distance compared to non-optimized baselines, validating its efficacy and scalability in complex, dynamic scenarios.

Keywords: Multi-agent systems, Distributed reinforcement learning, Consensus and reinforcement learning control
Abstract: This paper presents a safe multi-agent deep reinforcement learning framework for cooperative quadrotor formation flight based on bearing rigidity. A team of UAVs is required to navigate cluttered environments while preserving a desired formation shape and avoiding collisions. A rigidity-based bearing controller guarantees convergence to the desired shape up to global translation, uniform scaling and coordinated yaw (TSY). On top of this analytic layer, we embed a Multi-Agent Deep Deterministic Policy Gradient (MADDPG) architecture whose actors operate only in the TSY null-space, so learning affects the group motion but cannot inject formation distortion. Safety is enforced by a zeroing control barrier function (CBF) quadratic program that filters the nominal control into a safe joint velocity. Unlike conventional safe RL, we differentiate through the CBF–QP and train the centralized critic and decentralized actors on the executed safe actions, eliminating the train–test mismatch between nominal and filtered policies. Simulations in Gazebo with a bearing-rigid three–quadrotor formation show that the proposed method achieves higher success rate, faster and more consistent convergence, and significantly lower formation error than an RL+CBF baseline that acts in the full joint action space.

Keywords: Multi-agent systems, Learning methods for control, Distributed reinforcement learning
Abstract: In multi-agent Pursuit-Evasion (PE) scenarios, effective communication among pursuers is essential for successful coordination and capture efficiency. Traditional PE algorithms often face limitations due to fixed communication structures and inadequate adaptability to dynamic environments. To address these challenges, this study introduces a learning-based communication framework specifically designed for multi-target PE tasks. We enhance the existing Target-oriented Multi-Agent Communication and Cooperation (ToM2C) framework for multi-target PE scenarios by integrating an intensity-based filtering mechanism in place of its original Graph Neural Network (GNN) module. This filtering mechanism enables selective communication among pursuers based on confidence in target assignment predictions. Simulation results demonstrate significant improvements in both capture success rates and communication efficiency. Physical experiments validate sim-to-real transferability, confirming the effectiveness of the proposed approach.

Keywords: Multi-agent systems, Randomized algorithms in stochastic systems
Abstract: We study a class of distributionally robust games where agents are allowed to heterogeneously choose their risk aversion with respect to distributional shifts of the uncertainty. In our formulation, heterogeneous Wasserstein ball constraints on each distribution are enforced through a penalty function leveraging a Lagrangian formulation. We then formulate the distributionally robust game as a variational inequality problem, and show that under certain assumptions the original seemingly infinite-dimensional Nash equilibrium problem is equivalent to a multi-agent but finite-dimensional variational inequality problem with a strongly monotone mapping. Due to the inner maximization problem, it is however still challenging to calculate a distributionally robust Nash equilibrium. To this end, we design an approximate Nash equilibrium seeking algorithm and prove convergence of the average regret to a quantity that diminishes with the number of iterations, thus learning the desired equilibrium up to an a priori specified accuracy. Numerical simulations corroborate our theoretical findings.

Keywords: AI for smart cities, Low-altitude economy, Cyber-physical urban systems
Abstract: Object detection from Unmanned Aerial Vehicles (UAVs) is pivotal for the road monitoring task of smart city but faces severe challenges due to the prevalence of tiny objects. These targets suffer from spatial information decay, high-frequency feature submergence, and pixel misalignment within Deep Neural Networks (DNNs). To address these systemic bottlenecks, this paper proposes a Multi-stage Composite Feature enhancement FramEwork (MsCoFFe) for the current popular deep learning based UAV vision models. Unlike specific model patches, MsCoFFe is a general and plug-and-play framework designed to reinforce feature fidelity and alignment. It introduces the Feature Complementary Mapping (FCM) and Multi-Kernel Perception (MKP) modules in the backbone to preserve spatial details and enable multi-scale perception. Furthermore, it incorporates High-Frequency Perception (HFP) and Spatial Dependency Perception (SDP) modules in the neck network to amplify weak target signals and dynamically correct pixel shifts via cross-attention. The case study on the VisDrone2019 dataset demonstrate that integrating MsCoFFe into state-of-the-art deep learning object detectors, such as RT-DETR and DEIM, significantly improves detection robustness. Notably, the proposed MsCoFFe increases the AP50 of the DEIM model by 6.8%, validating its effectiveness in complex aerial surveillance scenarios with tiny objects.

Keywords: AI for smart cities, Smart city control and optimization, Cyber-physical urban systems
Abstract: Millimeter-wave radar is attractive for urban emergency response because it remains operative in darkness and visual obscurants, yet existing drone-detection systems trade 3D spatial resolution against temporal continuity. We present DmmD, a dual-mmWave-radar framework that combines a Multi-View Doppler Rectification Layer with an STC-Net based on 3D ConvLSTM. MVDRL aligns Doppler features from orthogonal views using geometric priors before fusion. Experiments on a synchronized dual-IWR6843 platform achieve 97.10 % AP, improve AP1 over CubeDN, and reduce mean localization error to 0.52 m. Barrier tests further show less than 1% point-cloud density reduction through visually opaque materials.

Keywords: Analytic design, Linear systems, Control of complex systems
Abstract: In the behavioral approach to systems and control, a system is characterized by the set of the trajectories, which is referred to as the behavior. Using this approach enables us to obtain solutions that are completely independent of mathematical expressions and to discuss them in a set-theoretical context. As considered in the standard control theory, one fundamental problem is whether a given control specification can be implemented for a particular plant. This issue has also been studied within the behavioral approach. In cases where the dynamics of a plant varies, it becomes important to determine the extent of acceptable changes. We formalized this problem as the simultaneous implementability problem, this means to consider what is a condition under which a single specification can be realized by using a single controller for two different plants. In this paper, we adopt an set-theoretical approach to examine the simultaneous implementability problem in the behavioral approach.

Keywords: Automotive system identification and modelling, Modeling, supervision, control and diagnosis of automotive systems, AI and learning-based control for automotive systems
Abstract: Automatic real-time tyre identification is crucial for improving vehicle performance, safety, and efficiency. This capability is valuable in racing applications, where it can support consistency checks and strategic decisions, and even more relevant in urban and aftermarket scenarios, where tyre information is often unavailable, and vehicle control systems could benefit from real-time adaptation. Despite its relevance, the literature mainly focuses on tyre usage monitoring. Furthermore, these approaches also reveal a trade-off between practicality and interpretability: model-based methods provide physically meaningful results but often require measurements that are rarely available in real-world vehicles, whereas machine learning methods exploit accessible vehicle signals and achieve high predictive performance, typically at the expense of interpretability. To address this gap, this paper presents and compares two real-time tyre classification strategies: a model-based method designed to rely on accessible vehicle measurements, and an interpretable learning-based approach. Their performance is assessed in both simulation and real-world experiments. While both methods achieve optimal performance in simulation, real-world variability and noise reduce the accuracy of the model-based approach. In contrast, the learning-based classifier maintains an F1-score of 96.5%, proving to be a practical and interpretable solution for real-time tyre recognition.

Keywords: Cognitive and emotional control/AI systems, arts and control, Cross-cultural aspects of engineering, Human-centric automation/AI Systems, and human agency
Abstract: Japanese culture places significant emphasis on emotionality alongside intellectual and logical aspects. This study examined the structure of initial trust in automation within the Japanese cultural context. Through exploratory and confirmatory factor analyses across three AI-enabled automation systems, the two-dimensional structure of initial trust, comprising cognition-based and affect-based initial trust, was supported. This finding is consistent with that observed in the Chinese context, although the specific items retained for each dimension were only partially aligned with those in the original Chinese scale. These results highlight the importance of distinguishing between cognition-based and affect-based trust in assessing initial trust in automation within both Chinese and Japanese cultural settings. Designers and practitioners should explicitly account for these two dimensions in the initial trust management of automation systems, thereby ensuring greater conceptual clarity and more accurate measurement.

Keywords: Control education laboratories, Industry-academia collaboration in control education, Internet based control education
Abstract: This paper presents an interactive virtual training system for the fault diagnosis and operation of twelve-phase rectifier generators, addressing the high cost and risks of physical training in control engineering education. Developed with Unity3D and Vue.js, the system enables principle learning, operational procedures, and fault injection in a simulated environment. A dedicated assessment module automatically evaluates trainee performance. The platform provides a safe, flexible, and effective tool for enhancing practical understanding and troubleshooting skills of complex marine electrical systems, demonstrating the significant value of virtual simulation technology in modern control education.

Keywords: Cyber-physical and human systems (CPHS), Human-centric automation/AI Systems, and human agency, System dynamics and control in CPHS
Abstract: This paper introduces a novel human skill evaluation framework that leverages multiobjective optimization to address the limitations of assessing human proficiency in dynamic, complex systems with unknown intentions. Previous methods struggle with multi-objective tasks, offer limited interpretability, or require extensive data. Our framework quantifies human skill by measuring the Euclidean distance from a human’s Key Performance Indicator (KPI) vector to the surface of Pareto optimal solutions. We explore various intention assumptions by selecting different points on the Pareto Front and evaluate their impact on skill assessment using manual parking maneuver simulations and demonstrate the framework’s real-time computability. The results highlight the influence of intention assumptions on skill evaluation and demonstrate the potential for a robust, interpretable, and adaptable approach for quantifying human skill.

Keywords: Cyber-physical and human systems (CPHS), Social computing, Game theories
Abstract: Trauma resuscitation is a clinical process for treating life-threatening physiological disorders in safety-critical environments, driven by the experience of healthcare workers (HCWs). Designing and optimizing quantifiable metrics that accurately capture HCW decisions may augment current resuscitation procedures with the potential to improve patient outcomes. This motivates our socio-technical formulation of trauma resuscitation as a distributed generalized Nash equilibrium (GNE)-seeking game with coupled inequality constraints. This method is optimized over a time-varying communication graph. We introduce novel insights from clinical experience to model HCWs behavior. This work facilitates the best possible resuscitation outcome given HCWs’ workloads, schedules, competencies, and limited resources.

Keywords: Cyber-physical and human systems (CPHS), System dynamics and control in CPHS, Human-centric automation/AI Systems, and human agency
Abstract: Human operators play critical roles in cyber--physical systems, yet control--theoretic models typically treat inter--subject variability as noise rather than as systematic patterns linked to individual characteristics. This article introduces a population mixed--effects framework for modeling human sensorimotor control that explicitly relates controller parameters to demographic and experiential covariates. Closed--loop identification experiments were conducted with 66 participants performing a single--axis target acquisition task, with the human modeled as a SISO controller and the plant as a kinematic integrator. Comparing PI, PID, and second--order structures, we find that the second--order model with a real zero consistently outperforms PI/PID, and that video game experience emerges as a particularly strong predictor of controller performance, with experienced players exhibiting faster response dynamics and improved tracking accuracy.

Keywords: Data centers and cloud computing, Decision making under uncertainty
Abstract: Integrating on-site renewable energy (RE) generation into data centers (DCs) offers a promising pathway toward energy sustainability. However, the inherent intermittency, volatility, and uncertainty of RE may expose DC energy systems to substantial risks of supply–demand imbalance. To address this challenge, this paper develops a stochastic energy management method for hydrogen-based geo-distributed data centers (HBGDCs). A remaining-time bucket mechanism is proposed to explicitly capture the temporal flexibility of DC workloads by dynamically tracking diminishing processing windows. Moreover, to handle forecast errors in renewable generation and workload arrivals, a receding-horizon scheduling framework is designed, in which a scenario-based two-stage stochastic optimization model is integrated. Numerical studies on a typical HBGDC system show that the proposed approach consistently improves operational efficiency under both normal and adversarial conditions, while being highly tolerant to forecast errors.

Keywords: Decision making under uncertainty
Abstract: This paper proposes a dynamic coalition game-theoretic framework for multispacecraft cooperative task allocation in adversarial environments with uncertain target priorities. The key innovation is an augmented time-varying characteristic function that integrates mission beneffts, execution costs, and transition penalties, with threat-adaptive weight mechanisms. We introduce an intelligence conffdence metric that dynamically evolves through observation, enabling adaptive target prioritization. The Shapley value allocation mechanism ensures fairness and stability while a utility maximization formulation with individual rationality constraints prevents coalition deviations. The system dynamically adjusts task assignments in response to changing threat levels, ensuring consistent performance over time by explicitly accounting for the costs of switching tasks and reorganizing teams.

Keywords: Differential or dynamic games, Applications of optimal control, Control barrier functions and state space constraints
Abstract: We study the reach-avoid problem, where a pursuer aims to capture an evader, targeting a target plane in three-dimensional space (3D) while avoiding exclusion zones. As there is no optimal control for situations involving exclusion zones, we propose using Reinforcement Learning (RL) to generalize the optimal control from scenarios without exclusion zones to those that include them. To guarantee that the pursuer does not enter the exclusion zones, we use Control Barrier Functions (CBF) as both a safety filter and as a measure of reward for the pursuer. We demonstrate the necessity of each proposed component within the framework by conducting an ablation study. Furthermore, the efficacy of the framework is validated through simulation against optimal control with CBF.

Keywords: Human-centric automation/AI Systems, and human agency, Decentralized economics/ecosystems (DeEco)
Abstract: Industry~5.0 is reshaping smart manufacturing toward human-centric production, where operators collaborate with AI systems and networked machines. In such environments, workstations, teams, and operators face different tasks and conditions, resulting in non-identically distributed (non-IID) data and heterogeneous expertise. These factors challenge centralized AI deployment and raise privacy, scalability, and robustness concerns. This paper proposes a human-centric peer-to-peer federated learning (P2P-FL) framework for collaborative skill transfer in Industry~5.0. Each worker or production cell is represented by an edge device that trains a local decision-support model and exchanges model parameters with socially or organizationally connected peers over a decentralized graph. To mitigate non-IID effects while preserving privacy and autonomy, we introduce trusted data sharing, where peers share only a small, controlled fraction of local data with selected neighbors. Using MNIST, CIFAR-10, CIFAR-100, and an industrial NEU surface-defect dataset with synthetic non-IID worker profiles, we compare FedAvg, FedProx, and P2P-FL with trusted sharing levels of 20% and 40%. Results show that modest sharing significantly improves final accuracy and macro-level performance while reducing client performance disparities. The findings highlight implications for human--AI collaboration, workforce upskilling, and AI assistants in Industry~5.0 smart manufacturing.

Keywords: Human-robot interaction, Medical and rehabilitation robotics, AI-powered robotics
Abstract: Surface electromyography (sEMG) is increasingly used in wearable human–robot interaction systems; however, inter-subject variability limits reliable transfer of gesture intent across users. This paper presents a style-invariant learning framework that enhances subject-independent sEMG-based gesture recognition without requiring subject identity labels. The method employs Instance Selective Whitening (ISW) for self-supervised pre-training to suppress subject-specific style from feature covariance, followed by supervised fine-tuning for gesture classification. Experiments on Ninapro DB1, DB2, and DB4 show improved accuracy and reduced cross-subject performance variance. The results suggest the potential of the proposed framework for adaptive sEMG-driven wearable HRI systems, while real-time robotic validation remains an important direction for future work.

Keywords: Industry-academia collaboration in control education, Control education laboratories, Control engineering curricula
Abstract: This paper presents the integration of a smartphone-based brake testing application, originally developed for periodic technical inspections (PTI) and expert practice, into an undergraduate course on vehicle motion. The TESTEK mobile application records vehicle acceleration using the internal sensors of Android devices and evaluates braking performance in accordance with UN ECE regulations, providing the mean fully developed deceleration (MFDD) and related indicators. The same application family has been deployed at all PTI stations in the Slovak Republic and has been validated against certified decelerometers, which makes its results suitable both for regulatory use and for education. We describe how real braking tests recorded by this application are reused in the subject Processes of Vehicle Motion as the basis for a kinematics assignment in which students analyse acceleration, velocity, distance and MFDD, and identify individual phases of the braking process. The assignment combines numerical integration, signal preprocessing and interpretation of results in the context of legislation. The proposed approach requires only low-cost hardware (a smartphone and, when needed, a generic OBD interface) yet provides students with authentic, industry-grade data and tools. We outline the course context, the design of the laboratory task, implementation experience and qualitative observations, and discuss planned extensions towards remote laboratories.

Keywords: Repositories for control education, Control education learning analytics, Control engineering curricula
Abstract: We present a small collection of conceptual multiple-choice questions (MCQs) on continuous-time LTI systems, designed for a second-year bachelor course on fundamentals of automatic control or dynamical systems. The questions target four recurrent misconceptions: (i) confusing internal (equilibrium) stability with external (BIBO) stability; (ii) believing that poles on the imaginary axis automatically imply bounded trajectories, irrespective of Jordan structure; (iii) assuming that repeated eigenvalues in state-space realizations necessarily cause loss of controllability or observability; and (iv) overlooking that equilibria and working points are solutions of linear algebraic equations whose existence and uniqueness depend on the column space and null space of the system matrix. The exercises are intended primarily as pen-and-paper MCQs (no calculators or computer algebra required), suitable for in-class formative assessment, written examinations, or as prompts for short oral discussions. The prerequisite learning outcomes (PLOs) include being able to solve linear systems of equations, compute eigenvalues and eigenvectors (and in some questions Jordan blocks), and interpret state-space models and BIBO stability. The assessed intended learning outcomes (ILOs) focus on distinguishing different notions of stability, relating boundedness to Jordan structure, diagnosing controllability/observability from input/output directions, and determining existence and uniqueness of equilibria. Annotated solutions explicitly address the targeted misconceptions and can be used as self-study material by students or as a discussion guide for instructors.

Keywords: Safety-critical and resilient systems, Human-centric automation/AI Systems, and human agency, Regulation, policy, and legal issues in control/AI
Abstract: This paper addresses a critical gap in the risk assessment of AI-enabled safety-critical systems. While these systems, where AI systems assist human operators, function as complex socio-technical systems, existing risk evaluation methods fail to account for the associated complex interaction between human, technical, and organizational components. Through a comparative analysis of system attributes from both socio-technical and AI-enabled systems and a review of current risk evaluation methods, we confirm the absence of explicit socio-technical considerations in standard risk expressions. To bridge this gap, we introduce a novel socio-technical alignment (STA) variable designed to be integrated into the traditional risk equation. This variable estimates the degree of harmonious interaction between the AI systems, human operators, and organizational processes. A case study on an AI-enabled liquid hydrogen (LH₂) bunkering system demonstrates the variable's relevance. By comparing a naive and a safeguarded system design, we illustrate how the STA-augmented expression captures socio-technical safety implications that traditional risk evaluation overlooks, providing a more system-theoretic basis for risk evaluation.

Keywords: Social networks for smart cities, Smart city security and resilience, Cyber-physical urban systems
Abstract: This paper investigates the resilient output consensus control for heterogeneous linear multi-agent systems (MASs) under matrix-weighted networks subject to denial-of-service (DoS) attacks. Matrix-valued interaction weights are employed to characterize the interdependencies among the multidimensional agent states. Differing from prior work on synchronous attacks, a more general scenario is considered where attacks independently and randomly disrupt individual interaction links, modeled by a Markov switching process. First, a fully distributed resilient estimator is proposed, enabling followers to estimate the leader state even under DoS attacks. Based on the estimator, a distributed control protocol is then developed to guarantee asymptotic output tracking in the mean-square sense for all followers. Finally, numerical simulations are conducted to validate the effectiveness of the proposed protocol.

Keywords: Demand response, Big data and machine learning applied to smart cities, Smart buildings and building automation
Abstract: Reinforcement learning (RL) offers a data-driven alternative to model-based control for building heating systems. However, most existing approaches focus solely on energy efficiency and thermal comfort, overlooking actuator degradation caused by frequent valve switching. This paper presents an RL-based control framework that jointly optimizes energy consumption, occupant comfort, and actuator longevity. Using a physically consistent neural network model trained on real data from the UMAR unit at the NEST building in Dübendorf, Switzerland, two RL algorithms—A2C and PPO—are evaluated under varying switching-penalty strategies and a smooth policy architecture (LipsNet). Results show that a PPO controller with a temperature-dependent switching penalty reduces valve cycles ten-fold while increasing energy use by only 7%. The LipsNet network further achieves comparable energy efficiency with four times fewer switching events. These findings demonstrate that incorporating hardware-aware constraints into RL training can extend actuator lifespan without compromising overall system performance.

Keywords: Energy management systems, Control and management of energy systems
Abstract: Thermal and performance control in modern CPUs faces a fundamental trade-off: maintaining thermal safety via DVFS (i.e., reducing frequency) limits performance, while overcooling wastes energy. We propose a cascade-like thermal management scheme that acts coordinately on frequency and coolant flow: the former counteracts millisecond-scale load variations to keep the chip safe, while the latter adapts heat removal on a slower time frame to reduce overcooling and associated energy waste. We also present a tuning strategy for the scheme, demonstrate its potential through simulations, and discuss technical viability in realistic settings such as data centres.

Keywords: Energy market, Electrical transmission systems, Energy management systems
Abstract: This study investigates the effect of the minimum transfer requirement of Line-Commutated Converter (LCC) HVDC systems on nodal price formation and congestion surplus in electricity markets. To systematically examine this characteristic, a two-bus DC Optimal Power Flow (OPF) model is proposed that explicitly incorporates both minimum and maximum power transfer limits. Because thyristor-valve-based LCC HVDC systems require a minimum level of power transfer through the converter, this operational characteristic imposes an asymmetric lower-bound constraint on power flow that does not arise in conventional AC transmission systems. Analytical results derived from the Lagrangian formulation demonstrate that when the minimum transfer requirement becomes binding, this lower-bound constraint directly influences the nodal price difference between regions. Consequently, even when power flows in the forward direction, the price differential may be reversed, giving rise to negative congestion surplus. These findings indicate that the minimum transfer requirement can materially affect nodal prices and market settlement outcomes. Simulation results corroborate the analytical findings, confirming that the minimum transfer requirement can cause congestion surplus to become negative under specific load conditions — an outcome that does not arise in a standard transmission-line model without this constraint. These results suggest that the operational characteristics of LCC HVDC may introduce variability into the settlement revenues of Financial Transmission Rights (FTRs). Accordingly, FTR market participants may benefit from explicitly accounting for the minimum transfer requirement when formulating bidding and hedging strategies, as it can alter both the direction and magnitude of nodal price differences and congestion surplus.

Keywords: Forecasting of power supply and demand, Energy management systems, Energy market
Abstract: Probabilistic load forecasting is required when operators must plan for both expected demand and forecast uncertainty. However, feature selection remains difficult for deep probabilistic models because their outputs describe lower and upper quantiles rather than a single point forecast. This study evaluates whether explainable artificial intelligence (XAI) attributions of model-implied predictive spread can support feature selection in probabilistic load forecasting. A Quantile CNN-LSTM is trained on ISO New England load, weather, market, and calendar data to produce 24-hour-ahead 90% prediction intervals. The lower and upper quantile forecasts are transformed into two explanation targets: an interval midpoint, representing demand magnitude, and an interval width, representing predictive spread. SHAP and Permutation Feature Importance (PFI) are used to rank features for each target. The rankings are tested through recursive feature ablation, tracking forecast error, interval width, and prediction-interval coverage. SHAP-based mean and width rankings, and PFI-based mean rankings, improve forecast accuracy by approximately 14–16% and move empirical coverage closer to the nominal 90% level. PFI-based width rankings do not provide the same benefit. Width-based feature selection did not outperform mean-based selection because the same demand and weather variables dominate both targets. The main contribution is therefore diagnostic: width attributions show whether features that drive demand magnitude also drive the model’s predictive spread, enabling feature selection to be evaluated directly from probabilistic model outputs rather than from a separate point-forecasting model.

Keywords: Forecasting of power supply and demand, Energy management systems, Power plant control
Abstract: 本研究は広域電力ネットワークの日先運用最適化手法を提案します。太陽光発電(PV)の発電と需要は信頼区間を持つ純電力需要予測プロファイルとして扱われ、需要予測プロファイルに対する運用パラメータの単調性を用いて、区間の内のいかなる実現にも最適性を維持する運用範囲を導出します。オペレーターには、各エリアにおける熱発電、バッテリー貯蔵、エリア間電力転送の上限と下限が提供されており、需要が信頼区間内でどのように振る舞っても最適な運用計画範囲を特定できます。本論文では、面積が少ないにもかかわらず電力網の構造を検討し、運用電力の単調性解析を行います。

Keywords: Power electronics, Electrical transmission systems
Abstract: This paper presents a methodology for retrieving the impedance parameters of subsystems within a radial power grid from global impedance measurements. The first stage involves filtering the contribution of topological parameters (e.g., connection cables) through a denoising autoencoder. Several network architectures were investigated and compared, including multilayer perceptrons, convolutional neural networks, and recurrent networks for both encoder and decoder structures. In the second stage, the parameters of the subsystems were identified by incorporating the relative proportion of each subsystem within the network into the machine learning algorithm. The proposed method was validated on a case study involving a wind farm equipped with power converters, where the identified parameters achieved an accuracy of up to 5%. The most effective configuration employed a multiplicative operation on the admittance feature map vectors. This study represents an initial step toward the development of aggregated power grid models derived solely from external measurements.

Keywords: Power electronics, Power systems stability, Energy management systems
Abstract: This paper proposes an improved observer-based backstepping control scheme for DC microgrids with constant power loads (CPLs). A prescribed-performance function (PPF) is employed to restrict the tracking error within predefined bounds, while an enhanced fixed-time control achieves a smaller settling-time bound. In addition, a sliding-mode disturbance observer (FT-SMDO) is developed to estimate the time-varying power flow of uncertain CPLs. To ensure optimal estimation performance and eliminate manual gain tuning, the Grey Wolf Optimizer (GWO) is utilized to automatically tune the FT-SMDO parameters. Simulation results demonstrate that the proposed method achieves faster voltage recovery, improved robustness, and superior overall performance compared with existing controllers.

Keywords: Power systems stability
Abstract: This paper proposes a mechanical analogy to provide an intuitive understanding of power system dynamics, especially for novices. Our approach is applicable to multi-machine systems and incorporates the high-fidelity Park's model. We demonstrate a comprehensive mapping where all state variables of the power system, including generators and loads, correspond to states in the analogy. This framework facilitates the understanding of complex nonlinear dynamics and is validated by establishing its rigorous correspondence with the system's energy function.

Keywords: Power systems stability, Distributed optimization for smart grids, Power plant control
Abstract: Load Frequency Control (LFC) is crucial for maintaining power system stability by restoring nominal frequency and balancing inter-area power flows after disturbances.The control‑disturbance interaction can be modeled as a linear zero-sum game within the H_infty control framework. While the Simultaneous Policy Update Algorithm (SPUA) has offered higher computational efficiency than the traditional double-loop method for linear zero-sum games, it relies on the Newton–Kantorovich conditions for convergence, making it highly dependent on specific initial conditions that are difficult to verify, especially in model-free settings.This paper employs a homotopy-based single-loop policy iteration method for solving linear zero-sum games arising in LFC. The method only requires an initial stabilizing controller, obtained through an iterative homotopy procedure, and avoids the need for system dynamics or a predefined initial matrix. As a result, it offers improved computational efficiency and reliable convergence. Simulation studies on a single-area power system demonstrate the method’s robustness and accuracy compared with SPUA approach.

Keywords: Power systems stability, Electrical transmission systems
Abstract: This paper performs a numerical analysis of transient stability in power systems using the convexity of the energy function and an analogy with mass-spring-damper systems. The Hessian of the energy function and its eigenvalues are interpreted as the spring constant matrix and spring strength, respectively. Numerical results demonstrate that increasing the spring constant matrix through parameter tuning of the VSG model enhances the system's transient stability. Furthermore, a positive correlation exists between the critical clearing time (CCT) during a ground fault and the stiffness. Using the analogy with the physical system, an intuitive interpretation is provided for the mechanism by which stronger springs increase CCT.

Keywords: Power systems stability, Electrical transmission systems, Control and management of energy systems
Abstract: We derive the nonlinear dynamical model of an AC power system consisting of multiple Synchronous Generators (SGs) and Distributed Energy Resources (DERs) interfaced with the grid by DC/AC power converters, in a generic meshed network topology that also incorporates the dynamical phenomena of the lines and the loads. In particular, the high-order nonlinear model is used for the SGs, while the converter units of the DERs are considered to operate in grid-forming mode, leading to dynamical modelling in the local rotating frame of each Generating Unit (GU), i.e. each SG and DER; thus facilitating the application of decentralised controllers. Based on the port-Hamiltonian nonlinear dynamical structure obtained for the complete power system, input-to-state stability (ISS) is analytically proven for the first time, as far as the authors know, when taking into account both SGs and grid-forming DC/AC converters in the power system model, considering also the sixth-order nonlinear model for the SGs. Furthermore, bounded integral controllers are designed for each GU that guarantee boundedness of the closed-loop system solution, without requiring any knowledge of system parameters, while additionally satisfying desired input constraints. A 4-bus power network is simulated to validate the ISS and boundedness properties of the developed dynamical model, as well as the input constraint satisfaction provided by the controllers.

Keywords: Power systems stability, Power electronics
Abstract: Maintaining phase-angle stability is crucial for grid-forming inverters in renewable-dominated power systems, particularly under severe disturbances and low short-circuit strength. To enhance stability resilience, the paper proposes a safety filter that shapes the active-power reference to keep the inverter–grid phase difference within a safe margin, thereby mitigating overcurrent and loss-of-synchronism risks. In contrast to traditional current-limiting or mode-switching methods, the proposed safety filter is implemented via a control barrier function and acts as a lightweight modification of the active-power reference while preserving the nominal control architecture during normal operation. Analytical results derived on a reduced-order model establish formal safety guarantees under bounded grid-angle jumps. Extensive reduced-order Monte Carlo simulations across diverse short-circuit scenarios validate reliable angle-margin preservation and the associated safety-intervention trade-off.

Keywords: Power systems stability, Power electronics, Power plant control
Abstract: This paper proposes a unified framework for modeling and large-signal stability analysis of grid-connected inverters. It demonstrates how the Takagi-Sugeno (TS) framework provides a rigorous theoretical foundation by representing three-phase inverter systems con- nected to the grid as a state- and input-dependent weighted combination of linear models. This paper details modeling and stability analysis, with particular emphasis on input-to-state stability (ISS), a structural requirement for inverter systems in which grid voltage fluctuations are uncontrollable inputs. To address the practical requirement of fully describing the inverter system’s operating range as defined by grid code specifications, this work presents a modeling method accompanied by LMI-based stability analysis in the large-signal domain—not merely the small-signal range as commonly found in the literature.

Keywords: Control and management of energy systems, Energy management systems, Control and optimization for sustainability and energy systems
Abstract: This paper addresses secondary-layer power management in DC microgrids with hybrid storage configurations, including partially uncontrollable fast devices such as supercapacitors. Unlike conventional approaches, we consider scenarios where fast storage outputs are dictated by primary-layer dynamics, while slower storage units track secondary-layer references. We propose a practical strategy that prevents the state-of-charge of uncontrollable devices from reaching extreme levels by temporarily operating them as energy buffers and introducing a control-mode signal to coordinate DC-bus stabilization and power tracking. The approach is implemented via Model Predictive Control, and simulations demonstrate that it ensures long-term microgrid stability while enhancing robustness and operational flexibility.

Keywords: Energy management systems, Forecasting of power supply and demand
Abstract: Optimizing greenhouse temperature to balance crop productivity and energy efficiency is a major challenge in protected agriculture. This work introduces an optimization framework that integrates climate, crop growth, and Energy Hub modeling. A bilevel GA–MILP (genetic algorithm - mixed integer linear programming) strategy is applied: the GA maximizes profit by calculating heating and cooling setpoints for adequate crop growth, while the MILP focuses on minimizing operational costs by energy scheduling. A simulated case study based on a Mediterranean greenhouse was used to evaluate the approach, achieving up to 43% cost savings compared to manually setting the temperature setpoints. Although this comes with a 8% reduction in revenue, the overall profit increases by 5%, representing a modest economic gain but a significant contribution to the sustainability of food production.

Keywords: Wind power, Control and management of energy systems
Abstract: Multi-kite Airborne Wind Energy systems harvest wind energy through several kites and tethers. While they are predicted to yield significantly higher power output than single-kite systems, they are also considered more complex, and practical real-world designs have yet to appear. We propose a novel multi-kite system in which the kites are constrained to orbit each other by tethers connecting their inner wingtips. The centrifugal stiffening in this arrangement results in a quasi-rigid rotor that transmits mechanical power to the ground-based generator by pulling out a Y-shaped tether. Such a system is modeled with high fidelity and controlled with simple means. This shows that the proposed architecture is less complex than commonly thought and has important advantages over previously proposed single-kite systems.

Keywords: Control and optimization for sustainability and energy systems, Wind power
Abstract: This paper investigates trajectory control of tethered aircraft flying on circular paths by exploiting motion primitives defined on a spherical surface. Using the motion primitives, we derive a longitudinal model of the aircraft and characterise the trim conditions required to maintain steady flight on a prescribed primitive. These trim conditions are then used as a feedforward law around which simple feedback controllers are designed. The simulation results show that combining trim-based feedforward and low-complexity feedback achieves accurate path-following and speed regulation, illustrating the potential of motion-primitive-based models for the guidance and control of tethered aircraft in airborne wind energy applications.

Keywords: Hydrogen systems for energy generation and storage, Control and management of energy systems, Energy storage systems
Abstract: This work proposes a constrained quadratic dynamic matrix control (QDMC) strategy to reduce hydrogen–oxygen cross-contamination in high-pressure alkaline electrolyzers, thus improving the purity of the supplied gases. To reduce gas contamination, the controller adjusts the opening of the two outlet valves based on the system pressure and the difference in liquid level between the two gas separation chambers. A quadratic dynamic matrix controller (QDMC) with constraints and multiple inputs and outputs (MIMO) is developed. The behavior of the closed-loop system under the proposed controller was assessed through simulation, employing a 25-state high-fidelity non-linear model. The simulation results show a hydrogen purity below 0.35% O2 under all scenarios.

Keywords: Hydrogen systems for energy generation and storage, Control and optimization for sustainability and energy systems, Demand response
Abstract: Green hydrogen production using Proton Exchange Membrane (PEM) electrolyzers can support the decarbonization of hard-to-electrify sectors. PEM electrolyzer systems can operate either off-grid using only renewable energy or in a grid-connected configuration that supplements renewables with grid electricity. While grid-connected operation improves flexibility and continuity of operation, the carbon intensity (CI) of the hydrogen produced depends on the time-varying emissions associated with the bulk grid. The economic performance of grid-connected systems also depends on how well operation is aligned with low electricity price periods, which requires short-term forecasting. This study develops a rolling horizon optimization (RHO) framework that incorporates updated SARIMA-based electricity price forecasts, renewable availability, and CI limits. A mixed-integer linear programming (MILP) model determines electrolyzer loading, compression, and storage decisions. Several representative operating days with different grid CI levels are examined. Without CI limits, production shifts toward low-price periods, resulting in average CI values between 3.8 and 6.6 kg CO₂e/kg H₂, depending on the CI of the grid electricity used. When CI limits are imposed, grid-only operation cannot satisfy the threshold on high CI days, whereas renewable availability enables low CI production near 1.2–1.6 kg CO₂e/kg H₂. On low CI days, constrained and unconstrained outcomes have negligible differences. These results show that meeting carbon-intensity requirements while maintaining economic performance requires scheduling strategies that account for both price variability and renewable availability.

Keywords: Hydrogen systems for energy generation and storage, Process modeling, identification, and estimation techniques, Energy storage systems
Abstract: Integrating liquefied natural gas (LNG) cold energy into hydrogen systems offers an opportunity to reduce cooling loads and improve process efficiency, yet its system-level benefits across production and storage stages remain underexplored. To address this gap, four hydrogen supply configurations combining two production routes (SMR and ATR) with two storage pathways (LOHC and NH3) were modeled, and exergy and techno-economic analyses were performed with and without LNG cold-energy integration. LNG cold energy reduced cooling and pre-conditioning demands in the storage section, providing moderate improvements in exergy efficiency and operating costs across all cases. LOHC-based systems achieved the highest efficiencies (91–92%) and the lowest levelized hydrogen costs (1.99–2.38 /kg), with the SMR–LOHC configuration exhibiting the most favorable performance. In contrast, NH3-based systems showed lower efficiencies (81–83%) and higher costs (3.26–3.68 /kg) due to additional energy demands associated with high-pressure synthesis and multi-stage compression. This study offers a quantitative assessment of LNG cold-energy use across both production and storage stages and demonstrates its potential to enhance the efficiency and economic viability of LNG-based hydrogen systems, while clarifying system-level trade-offs between LOHC and NH3 storage routes.

Keywords: Hydropower
Abstract: Monitoring the energy efficiency of hydropower units is critical for production optimisation and predictive maintenance, but direct measurement through thermodynamic tests is costly and seldom performed. Continuous estimation from standard operational data is therefore desirable, yet challenging due to the absence of direct net head instrumentation and to flow-dependent non-stationary noise on industrial sensors. This paper proposes a ``grey-box'' methodology in which three physics-based dynamic models for the net head (Pressure-Based, Surge-Tank-Based, and Upstream-Reservoir-Based) are coupled with Adaptive Cubature Kalman Filters (ACKF) and Smoothers (ACRTSS). Process-noise non-stationarity is handled by a sliding-window variance estimator applied directly on the noisy input signals. Validation on a high-head plant with real industrial data shows that the proposed dynamic smoother reduces the RMSE against thermodynamic references by approximately 30% and improves temporal stability by a factor of five compared with filtered static methods.

Keywords: Nuclear power, Power plant control
Abstract: To mitigate the high operator workload and operational risks associated with manual control rod operation during the start-up stage of High-Temperature Gas-Cooled Reactors (HTGRs), this paper proposes an automated power control method. The start-up process is divided into two power ranges: 0–30% and 30–50% of Rated Full Power (RFP). The operation of control rod is automated by presetting parameters such as the operation sequence, position limits, step size, and interval time. In the 0–30% RFP stage, the flow rates of the primary circuit coolant and the secondary circuit coolant are fixed. In the 30–50% RFP stage, a linear ramp-up strategy for feedwater flow rate is implemented to effectively suppress the excessively steam temperature and ensure a stable steam temperature increase, while primary helium flow rate remains unchanged. Simulation results demonstrate that the proposed method achieves stable power increase and confirms its control performance and operational safety. Furthermore, this study analyzes the influence of negative temperature feedback on reactor power and examines the stabilizing effect of feedwater regulation on steam temperature. The findings provide the practical insights for the automatic control of start-up stage of HTGRs.

Keywords: Thermal systems modelling, Control and optimization for sustainability and energy systems, Energy management systems
Abstract: Decarbonizing the global energy supply requires more efficient heating and cooling systems. Model predictive control enhances the operation of cooling and heating systems but depends on accurate system models, often based on control volumes. We present an automated framework including time discretization to generate model predictive controllers for such models. To ensure scalability, a primal decomposition exploiting the model structure is applied. The approach is validated on an underground heating system with varying numbers of states, demonstrating the primal decomposition’s advantage regarding scalability.

Keywords: Guidance, navigation and control for AVs, Automatic control, optimization, real-time operations in transportation, Control architectures in automotive control
Abstract: This paper addresses bearing-only algorithms for solving the Fermat-Weber Location Problem (FWLP) with a unicycle agent. Unlike existing FWLP solutions for single- or double-integrator agents, our approach accounts for the nonholonomic constraints of wheeled robots. We first develop a bearing-only control law for the case with stationary beacons. Next, we consider saturated control inputs and propose a corresponding bearing-only control law. Finally, we address moving beacons with constant velocities and develop a control law that enables the unicycle agent to track the moving Fermat–Weber point. Both simulations and experiments are provided to demonstrate the effectiveness of the proposed methods.

Keywords: Multi-vehicle systems, Intelligent transportation systems, Trajectory tracking and path following for AVs
Abstract: Cutting-corner behavior and loss of string stability are two principal concerns on platoon performance over curved roads. Because vehicle following governs how a platoon responds to curvature, it directly determines the significance of cutting-corner effects. Inspired by Newell’s car-following model, we propose a curved-road following method that uses the predecessor’s time-delayed state as the reference for each follower, enabling accurate tracking while avoiding cutting-corner behavior. Building on this method, we design a model predictive control (MPC) scheme that avoids cutting corners while maintaining the desired inter-vehicle spacing. With appropriately selected controller parameters, the closed-loop platoon preserves string stability. Simulation results validate the proposed following method and show that the MPC controller both prevents cutting-corner behavior and preserves string stability along the platoon.

Keywords: Nonlinear adaptive control, Neural and fuzzy adaptive control, Learning methods for control
Abstract: To address the control challenges posed by the strong nonlinearity and parameter uncertainty in the roll channel of canard-guided dual-spin projectiles, a fixed-time tracking control method based on radial basis function neural networks is proposed. Initially, a seven-degree-of-freedom coupled rigid-body dynamics model for the dual-spin projectile was developed, treating aerodynamic parameter uncertainties as lumped disturbances. The model was then decoupled into roll channel and pitch/yaw channel dynamics subsystems using time-scale separation. Radial basis function neural networks were employed to precisely approximate the model uncertainties. Moreover, filters were introduced to compute the virtual derivatives, effectively preventing the common issue of "derivative explosion" in traditional control systems. The designed controller integrates roll angle tracking error feedback with lumped disturbance estimation feedforward, aiming to achieve fixed-time convergence and enhance the system's convergence speed and robustness, thereby ensuring precise roll angle tracking control. Using the Lyapunov method, the uniform ultimate bounded stability of the closed-loop system was demonstrated. Simulation results indicate that under conditions of aerodynamic parameter perturbation with a frequency of 1000 Hz and amplitude deviation of ±30%, the method can achieve an average roll angle tracking error of no more than 0.1 degrees, exhibiting excellent maneuver command tracking precision and robustness.

Keywords: Nonlinear adaptive control, Nonlinear system identification, Learning methods for control
Abstract: Adaptive control has been extensively used to estimate constant unknown parameters in uncertain nonlinear dynamical systems and to exploit those estimates to improve tracking performance. Memory regressor extension (MRE) methods leverage accumulated input–output data to relax excitation requirements, with full-data MRE integrating the entire history of regressors to drive parameter updates. Alternatively, forgetting-based MRE introduces selective data discounting to retain the benefits of stored information while improving robustness to disturbances. Forgetting-based estimation methods achieve this by constructing an information matrix (IM), i.e., an integral regressor matrix whose stored data is strategically discounted to accommodate changes in the dynamics. Traditional exponential forgetting applies a uniform decay across the entire regressor space, which can cause estimator windup under poor persistence of excitation (PE), where the IM becomes positive semi-definite, and the parameter estimates deteriorate over time. In contrast, directional forgetting (DF) discounts data only along the subspaces spanned by new information in the regressor. Although existing DF approaches, including orthogonal and oblique projection methods, successfully prevent estimator windup, they are often limited to first-order dynamics, assume exact knowledge of the system, and fail to address closed-loop tracking, limiting their applicability. This paper develops a nonlinear adaptive control scheme that incorporates oblique DF into a closed-loop design for uncertain Euler–Lagrange systems, achieving both kinematic tracking and parameter estimation. Integral data-driven regressors and input vectors are used to avoid computing second-order derivatives. A Lyapunov-based analysis establishes global exponential convergence of both tracking and parameter estimation errors under the PE condition. Numerical simulations of a two-degree-of-freedom robotic system validate the developed method, demonstrating satisfactory tracking performance and reliable estimation of constant unknown parameters.

Keywords: Nonlinear adaptive control, Stochastic adaptive control
Abstract: This article investigates the adaptive backstepping fault-tolerant control (FTC) problem for uncertain large-scale time-delay systems subject to input saturations. By establishing a technical lemma, the growth assumption imposed on the delayed interactions is successfully removed. Then, an adaptive FTC scheme is presented, which is capable of accommodating the stochastic intermittent failures of multiple saturated actuators. With the aid of a Lyapunov-Krasovskii functional, it is proven that all the closed-loop signals remain globally ultimately bounded in probability. Also, it is established that the tracking error can be reduced by tuning design parameters in a explicitly manner.

Keywords: Perception and filtering in marine systems, Sensors and actuators in marine systems, Decision and support in marine systems
Abstract: This work addresses multitask recognition of the active engine (M1–M5) and its operating state from underwater audio. We compare four shared feature-extraction networks, here termed backbones, namely BiLSTM+attention, GRU+attention, a temporal Transformer, and ResNet-50 on spectrograms, all coupled to conditional state heads. Preprocessing uses 0.5 s windows of 64-bin log-mel spectrograms, z-score normalization, and light augmentation (random gain, Gaussian noise, and SpecAugment). Experiments are conducted on the single-engine subset of Wolfset, with evaluation at segment and file levels. Among the reference models, GRU and Transformer reach file-level F1 of 1.00 for engine and up to 0.68 for state. Motivated by these results, we propose a sub-spectrogram GRU variant; with B=8, it yields the best trade-off (mean F1 = 0.800; file-F1: engine = 1.00, state = 0.74). Removing augmentation substantially degrades state recognition (file-F1 0.74→0.47). On a Tesla T4 GPU, end-to-end inference over a complete file under the adopted windowing required 83–113 s with memory usage < 225 MB, supporting batch or near-online monitoring rather than strict real-time deployment.

Keywords: Power and propulsion in marine systems, Modelling, identification and control in marine systems, Marine renewable energy systems
Abstract: Ship electrification is a major enabler for zero-emission shipping and the use of alternative fuels and power sources. However, they contribute to higher complexity of energy systems, which leads to suboptimal operation for conventional rule-based control. Alternatively, advanced control can take the available knowledge about the vessel and its operation into account. This paper presents a nonlinear multi-objective Model Predictive Control approach for a hybrid-electric vessel energy system to enhance energy efficiency. In a simulation study, the controller shows the potential to reduce fuel consumption by 2.5 %.

Keywords: Rail transportation modelling and control systems, Planning, management and security in transportation, Modeling and simulation of transportation systems
Abstract: This paper provides a control-engineering reading of railway infrastructure monitoring, formally stating the underlying maintenance problem as a partially observed sequential decision problem and reviewing, through this lens, the available observables, the methodological pipelines, and the bottlenecks that prevent closing the loop in practice. While modern sensors achieve sufficient observability, the integration of heterogeneous data into a closed prescriptive loop remains fragmented. We identify three structural challenges: multi-scale temporal fusion, the performance-explainability trade-off, and the lack of longitudinal benchmarks for sequential decision-making. On this basis, we outline a roadmap toward hybrid supervision systems combining physics-based estimators, probabilistic prognosis and constrained decision policies.

Keywords: Rail transportation modelling and control systems, Trajectory tracking and path following for AVs, Autonomous vehicles
Abstract: We present a model-based methodology to synthesize velocity controllers for individual Urbanloop pods, which are autonomous railway-based vehicles. They are designed for energy-efficient, low-cost, rapid, and seamless urban transport. First, we derive a physics-based pod dynamical model and rigorously reveal that it exhibits two time scales. We then leverage singular perturbation methods combined with recent contraction theory tools to design the controller, guaranteeing that the pod velocity tracks the given reference velocity profile. This controller combines a contractive output-feedback component with a reference-inducing feedforward term. We prove that the trajectories of the original, full-order model exponentially converge to the reference trajectory up to an error proportional to the time-scale separation parameter. Finally, numerical simulations illustrate the relevance of the approach.

Keywords: Resilient networked control systems, Cyber security networked control, Consensus
Abstract: This paper addresses the safety control problem of a self-organized swarm in environments with obstacles and adversaries. To mitigate adversarial impacts, a reputation mechanism is introduced for both leaderless and virtual-leader scenarios to quantify mutual trust among agents. This mechanism integrates local behavioral assessments with neighbors' reputations, allowing agents with low reputations to be regarded as potentially malicious. Such malicious agents are then isolated through communication weight adjustments at the cyber layer and repulsive potential fields at the physical layer. The distributed safety control laws are designed to ensure self-organizing characteristics and collision-free maneuvers. Simulation results demonstrate that the proposed approach effectively preserves self-organized swarm behavior and guarantees safety despite the coexistence of obstacles and adversaries.

Keywords: Space exploration and transportation
Abstract: Planetary rover exploration is attracting renewed interest with several upcoming space missions to the Moon and Mars. However, a substantial amount of data from prior missions remain underutilized for path planning and autonomous navigation research. As a result, there is a lack of space mission-based planetary datasets, standardized benchmarks, and evaluation protocols. In this paper, we take a step towards coordinating these three research directions in the context of planetary rover path planning. We propose two large planetary datasets, MarsPlanBench and MoonPlanBench, derived from high-resolution digital terrain images of Mars and the Moon. In addition, we set up classic and learned path planning algorithms, in a unified framework, and evaluate them on our proposed datasets using a popular path planning benchmark. Through comprehensive experiments, we report new insights on the performance of representative planning algorithms on planetary terrains, for the first time to the best of our knowledge. Our results show that classic methods can achieve up to 100% global path planning success rates on average across challenging terrains such as Moon's north and south poles. Conversely, learning-based models, although showing promising results in less complex environments, still struggle to generalize to planetary domains. Code and datasets available at: https://github.com/mchancan/PlanetaryPathBench.

Keywords: Space exploration and transportation, Aerospace mission control and operations, Guidance, navigation and control of aircraft and spacecraft
Abstract: This paper presents an optimization-based framework for designing multiple asteroid flyby missions in the inner Solar System. The core of the methodology leverages Venus gravity assists to place the spacecraft on controlled resonant orbits, enabling the construction of complex flyby sequences. We formulate the trajectory design as a two-stage optimization problem: first, a geometric pre-selection identifies candidate asteroids based on resonant orbit manifolds; second, a global-local optimization technique minimizes the total velocity increment (Delta v) while satisfying constraints on gravity-assist turn angles and launch energy. Numerical results demonstrate the method’s efficacy, generating fuel-efficient tours from a 2029 launch that include up to seven asteroid flybys with a launch Delta v under 3.6 km/s. The proposed approach demonstrates that resonant flyby sequences are highly competitive with direct transfers, often reducing propellant require

Keywords: Trajectory and path planning for AVs, Autonomous vehicles
Abstract: The problem of trajectory planning in stochastic, dynamic environments is inves tigated, with an emphasis on formulating efficient collision avoidance constraints. Black-box predictors provide an estimate of the stochastic obstacles’ state and the uncertainty of this estimate is quantified off-line via the statistical tool of Conformal Prediction. The resulting quantification is combined with elements of convex geometry, leading to the construction of the unsafe sets, regions which the obstacles, admitting polytopic representations, may occupy. The unsafe sets preserve the properties of compactness and convexity. Thus the safety constraints involving them and an agent with polytopic representation, may be efficiently formulated utilizing the Hyperplane Separation Theorem. The proposed optimization-based trajectory planning algorithm provides probabilistic collision avoidance and recursive feasibility guarantees, over finite time horizon, via progressive tightening of the unsafe sets. Its efficacy is demonstrated in the context of autonomous parking scenarios.

Keywords: Trajectory and path planning for AVs, Autonomous vehicles, Cooperative navigation
Abstract: To enable an efficient interaction of non-communicating mobile robots in collision avoidance scenarios, we present a novel combined trajectory planning and prediction algorithm. Inverse optimal control is used to estimate unknown goal states of all robots based on observed past trajectories. Each robot also takes the perspective of other robots in considering self-prediction and solves a joint prediction problem using the estimated goal states. The resulting predictions are then considered for planning. Simulation results of scenarios with 2-8 robots show that the median of the durations until all vehicles reach their goals is 9.8 % faster compared to planning with constant acceleration based estimated goal states. Moreover, the proposed approach never leads to the solver being unable to find a solution to the planning or prediction problem.

Keywords: Trajectory tracking and path following for AVs, Guidance, navigation and control for AVs, Autonomous vehicles
Abstract: Autonomous racing provides a valuable testbed for evaluating controllers in high-speed, traction-limit conditions. On scaled platforms, however, limited sensing and computation restrict the use of Model Predictive Control, motivating lightweight controllers that still capture nonlinear tire effects. This paper proposes the Adaptive Understeer Gradient (AUG) controller, a closed-form steering law that converts L1 guidance-based desired lateral acceleration into steering command while adaptively reflecting tire nonlinearity. It requires only a few parameters, no lookup tables, and can be tuned in real-time. Experiments in simulation and real-world F1TENTH racing show that AUG significantly reduces cross-track error and lap time compared to Pure Pursuit, while requiring far less tuning effort than existing dynamics-aware controllers. The code is available at: https://github.com/skcworld/controller.

Keywords: Trajectory tracking and path following for AVs, Guidance, navigation and control for AVs, Marine system guidance, navigation and control
Abstract: Path following should be evaluated along the path, not in time. We present a metric for comparison of path following using the line integral of the absolute value of the cross-track error along the desired track. The metric, which we term IAX, and its variants, ensure fair comparison regardless of the speed of progression along the path. We demonstrate in two cases that IAX is beneficial over the integral of absolute error (IAE) for such scenarios, and also provides a spatial interpretation in the plot. A toolbox is provided for ease of calculation of the proposed metric.

Keywords: Transportation logistics
Abstract: This systematic survey reviews the design and operational efficiency of unit-load warehouses utilizing multiple pickup and deposit (P/D) points. We analyze the evolution of facility layouts from traditional parallel aisles to non-traditional configurations, including Fishbone and Flying-V designs, specifically within multi-dock environments. The study categorizes literature based on storage policies, command cycles, and dock arrangements to evaluate their collective impact on travel distance. By synthesizing findings on optimal dock placement, this paper identifies critical research gaps and provides design guidelines for maximizing performance in modern logistics facilities.

Keywords: Uncertain systems, Linear parameter-varying systems, Linear time-delay systems
Abstract: Modern control design for dynamical systems must account for system uncertainties, including both static and dynamic ones. The primary challenge is to develop computationally efficient methods that can reliably capture the resulting stochastic system behavior. This paper proposes a novel and efficient uncertainty quantification method to represent a stochastic dynamical system through its mean and covariance trajectories. The approach models dynamic disturbances as a Gaussian Process, which is then reformulated as a Stochastic Differential Equation (SDE) to avoid the high computational cost of traditional Karhunen-Loève expansions. By combining this SDE representation with a surrogate model based on intrusive polynomial chaos expansion, we can analytically derive the mean and covariance dynamics for the system. This allows for a fast and accurate propagation of both static (parametric and timing) and dynamic uncertainties through the system model, making it suitable for advanced control design and online applications like model predictive control. The approach is illustrated by an application from longitudinal vehicle motion control.

Keywords: Electric and solar vehicles, Adaptive and robust control of automotive systems, Modeling, supervision, control and diagnosis of automotive systems
Abstract: Electric vehicles (EVs) incorporate regenerative braking mechanisms that enhance energy efficiency through the recovery of kinetic energy; however, rapid torque response of electric motors and transients during brake blending can increase longitudinal jerk and degrade passenger comfort. This paper proposes a comfort-aware brake blending strategy using an adaptive neuro fuzzy inference system that outputs a regenerative braking ratio (Z in [0,1]) derived from vehicle speed, battery SOC, and a motion- sickness indicator based on the Motion- Sickness Dose Value (MSDV). The controller was trained in MATLAB/Simulink using simulated braking scenarios, with training targets designed to balance SOC recovery and MSDV growth. A case of deceleration from 80~km/h to 40~km/h demonstrated that the proposed controller reduced MSDV to a greater extent than PID and fuzzy control while preserving regenerative energy recovery. These preliminary results suggest that dynamically adjusting regenerative braking intensity based on input variable changes enables efficient energy recovery while effectively suppressing passenger motion sickness.

Keywords: Uncertain systems, Probabilistic robustness
Abstract: The goal of this paper is to approximate the worst-case transient performance of uncertain linear time-invariant systems, subject to both L2-bounded input signals and known disturbances, e.g., reference tracking commands. System uncertainties are described through real-valued random variables with a known probability distribution. The worst-case performance analysis is formulated as a parametric Riccati differential equation, which is approximately solved using polynomial chaos expansion. The objective is to estimate a bound on the Euclidean norm of the system output at a given time. The effectiveness of the approach is demonstrated on the example of a spacecraft attitude and orbit control system.

Keywords: High-performance motion control systems, Robot perception and sensing, Aerial, field, and marine robotics
Abstract: This study proposes a confidence-aware vision-based control architecture that directly generates control inputs from visual errors. These visual errors are processed using a Python-based image processing pipeline. Typically, vision-based gimbal control delivers geometric detection results directly to the controller. However, in real-world environments, the reliability of visual information varies drastically over time, owing to changes in illumination, occlusions, or frame losses. In this study, OpenCV-based region of interest extraction, combined with a lightweight Tiny-CNN model,is used to estimate the confidence of detection results. The estimated confidence is incorporated into the control input via a soft-gating mechanism. The proposed structure continuously adjusts the input magnitude in undetected or low-confidence intervals, minimizing discontinuous transmission of control signals. The visual error and confidence data generated in the Python environment are integrated with a MATLAB/Simulink control system via a file-based interface. The controller is implemented in two configurations: (i) an ANFIS-based parallel compensation structure, and (ii) a pure ANFIS structure. The experimental results demonstrate that the proposed confidence-aware input modulation reduces response deviations under disturbance conditions and improves input stability.

Keywords: Human-centric automation/AI Systems, and human agency
Abstract: Human-AI collaboration raises important questions about agency and control, particularly as Large Language Models (LLMs) are increasingly integrated into knowledge work. This late-breaking work paper reports preliminary findings from a qualitative study investigating qualitative researchers’ perceptions of LLMs in their research practices, particularly focusing on how LLMs’ agency affects their acceptance of the technology. Results indicate that delegating core analytical work to LLMs raises concerns about the loss of control over the research process. The findings suggest that LLM design should prioritize flexible, user-controlled support, rather than the complete automation of tasks that are considered central to research practices.

Keywords: Human-robot interaction, Human AI integration, AI-powered robotics
Abstract: The digital twin of a 6-axis robot created in this project will serve as a basis for future research projects, particularly in the areas of machine learning, the explainability of AI, and the use of innovative human-technology interactions like Mixed Reality. The suitability of Unity as a platform for the holistic implementation of digital twins will also be explored. A physically representative model of the real twin will be created and communication between the twins will be implemented. The motivation for this is that Unity provides all the necessary tools to create a 3D visualization of the robot and its environment and also provides means for integrating complex programs via linked C# scripts.

For simulation purposes, the digital twin will be given independent control. Unlike many existing applications, the robot will not only be driven by simple movement equations that shift components, but also by motor forces in its joints. The physics engine included in Unity will be used for this purpose without additional modules. In addition, the digital twin should be able to track the movements of its real-world counterpart through communication.

Keywords: Human-robot interaction, Human machine cooperation & integration, Robot perception and sensing
Abstract: This paper presents an experimental design for constructing a multimodal dataset to analyze user engagement in human–robot interaction (HRI). Prior studies have mainly relied on observable behavioral cues, with limited frameworks integrating physiological signals. We therefore propose a structured data-collection protocol to build a multimodal dataset that includes wearable physiological signals, behavioral data, and self-report measures under different levels of task complexity defined in this experiment.

Keywords: Industrial artificial intelligence, Human-technology integration in manufacturing, Cyber-physical production systems
Abstract: As we strive to redefine future manufacturing systems that embrace Artificial Intelligence, we have started to think about ethical and sustainable ways to integrate this technology to ensure it benefits the human. Human-centric Industrial AI has started to take different shapes in the scientific community, but questions on what we mean by having the human do the creative tasks is not yet well defined. This paper makes an attempt to define what human augmentation means through a Human-Industrial AI Alignment framework, providing concrete ideas of how it could be implemented and discusses the feasibility of human creativity in an AI-powered automated environment.

Keywords: Maintenance engineering, management and services, Cyber-physical production systems, Smart production and logistics in manufacturing
Abstract: The digitalization of manufacturing systems driven by Industry 4.0 has increased importance of the unambiguous exchange of measurement data. As measurement data are expected to be processed more frequently by machines and algorithms, the Digital (International) System of Units (D‑SI) was proposed as a machine‑readable, SI‑based metadata scheme for describing measurement data. Given the central role of time‑series measurement data in manufacturing systems, this study proposes a specification‑based validation framework for D‑SI‑compliant measurement data. In the proposed framework, specifications are written as mathematical logical expressions, and validation is performed by evaluating whether the data satisfies these expressions. To illustrate this approach, we formalize three representative and fundamental types of specifications—unit consistency, occurrence of a specific measurement value, and temporal relationships between physical quantities.

Keywords: Manufacturing plant simulation, control and optimization, Industry X.0 for production and logistics, Maintenance engineering, management and services
Abstract: Industry 5.0 requires prescriptive Digital Twins (DTs) that exhibit stochastic stability against model uncertainty. This paper investigates the stochastic stability of an optimized operational control policy under a rigorous stress-test protocol where reliability parameters are subjected to ±20% model uncertainty. We formalize a "Stochastic Predictive Threshold Controller" (Deterministic Substitution) to absorb stochastic variance through managed preventive stops. Validated via N=30 Monte Carlo replications, results demonstrate a robust mean profit of €3,011.99 and a 100% service level despite high entropy. Crucially, we show that sustainability-driven heuristics discover stability margins that protect system bottlenecks, providing the statistical assurance required for deploying prescriptive DTs in volatile industrial networks.

Keywords: Teleoperation, Human-robot interaction, Robotic learning and adaptation
Abstract: This study investigates the feasibility of automating syringe disposal using a bimanual robotic system to minimize the manual handling of contaminated sharps. Using an imitation learning policy trained on 50 demonstration episodes, the system achieved a 76% end-to-end success rate. Analysis identified the needle detachment phase as the primary bottleneck due to alignment errors, while other stages remained highly reliable. These results demonstrate the potential for robotic automation in medical waste management using limited data. Furthermore, this research provides a foundation for ensuring the safety of healthcare professionals and preventing secondary infections through the development of robust robotic disposal systems.

Keywords: Data-driven control theory
Abstract: In this work, we present a novel data-driven formulation to solve the pole placement problem with or without the additional requirement of disturbance decoupling. The proposed approach is based on determining a data-driven solution to constrained Sylvester equations. Exploiting these equations, we also study the effect of process noise and develop an approach to obtain a pole placement design that is robust to such noise. Finally, the proposed methods are illustrated by means of a numerical example.

Keywords: Data-driven control theory
Abstract: This paper proposes a data-driven design method for a fixed-structure finite impulse response feedforward controller for an unknown stable discrete-time SISO LTI plant, assuming that only finite-length input-output data are available. The proposed method explicitly provides an upper bound on the finite-horizon worst-case tracking error over a prescribed finite set of representative reference signals. Using a Hankel-matrix-based trajectory representation derived from Willems' fundamental lemma, input-output trajectories consistent with the finite data are represented, and the minimization of the worst-case tracking-error bound over the prescribed reference set is formulated as a second-order cone program. This enables an offline one-shot design of a feedforward controller with a certified error bound for the prescribed finite reference set, without explicit model identification. Numerical simulations demonstrate that the proposed method constructs a feedforward controller with a finite-data-based error bound and achieves favorable reference-tracking performance.

Keywords: Data-driven control theory
Abstract: We propose a data-driven predictive control scheme for nonlinear systems with closed-loop stability guarantees. Using tools from Koopman operator theory and behavioral systems theory, we construct a data-driven lifted representation that enables the use of linear predictive control techniques for nonlinear dynamics. We show that, under suitable assumptions, exponential stability of the closed-loop lifted system implies exponential stability of the closed-loop nonlinear system. To address noisy measurements, the approach incorporates regularization and truncated singular value decompositions of the data matrices. Numerical experiments illustrate that the proposed framework achieves reliable trajectory tracking even in noisy settings and performs consistently across Hankel, Page, and Trajectory data matrices.

Keywords: Data-driven control theory
Abstract: Maglev trains, a leading technology for next-generation rail transit, offer a promising path to higher operating speeds. However, model-based control methods are challenged by model inaccuracies, while conventional PID control requires extensive parameter tuning. This paper proposes a direct data-driven strategy that bypasses modelling process. Specifically, the data-enabled policy optimization method is designed to use real-time train operation data to learn and adjust the control policy, thereby ensuring train adaptivity and stability. To account for mechanical and electromagnetic variations in the maglev modules, a regularization term is incorporated to the cost function, thereby rendering it robust to complex disturbances, even within a constrained adjustment space. Furthermore, the algorithm can be implemented recursively and hence is computationally efficient. Simulations demonstrate the effectiveness of the proposed approach.

Keywords: Data-driven control theory, Active learning and experiment design, Consensus and reinforcement learning control
Abstract: Compressed sensing enables sparse sampling but relies on generic bases and random measurements, limiting efficiency and reconstruction quality. Optimal sensor placement uses historcal data to design tailored sampling patterns, yet its fixed, linear bases cannot adapt to nonlinear or sample-specific variations. Generative model-based compressed sensing improves reconstruction using deep generative priors but still employs suboptimal random sampling. We propose an adaptive sparse sensing framework that couples a variational autoencoder prior with reinforcement learning to select measurements sequentially. Experiments show that this approach outperforms CS, OSP, and Generative model-based reconstruction from sparse measurements.

Keywords: Data-driven control theory, Adaptive observer design, Learning methods for control
Abstract: This paper investigates the linear output regulation problem with both the exosystem and the plant fully unknown. A data-driven regulator is proposed to achieve asymptotic regulation and closed-loop stability without performing model identification. The method constructs a nominal approximate internal model and filters of input and outputs, thereby yielding a stabilizable cascaded nominal system whose states are available. For this nominal system, a stabilizing law is derived from an offline dataset that has been acquired from the plant during experiments, such that the system states exponentially converge to a subspace. An identifier in discrete-time is, then, implemented to correct the internal model and update the stabilizing law; as a result, the regulation error can be steered to zero asymptotically under some persistent excitation conditions.

Keywords: Multi-agent systems, Resilient networked control systems, Distributed control and estimation
Abstract: This paper presents a geometric control framework on the Lie group SO(3) for 3D source-seeking by robots with first-order attitude dynamics and constant translational speed. By working directly on SO(3), the approach avoids Euler-angle singularities and quaternion ambiguities, providing a unique, intrinsic representation of orientation. We design a proportional feed-forward controller that ensures exponential alignment of each agent to an estimated ascending direction toward a 3D scalar field source. The controller adapts to bounded unknown variations and preserves well-posed swarm formations. Numerical simulations demonstrate the effectiveness of the method, with all code provided open source for reproducibility.

Keywords: Multi-agent systems, Control of networks, Control over networks
Abstract: This paper presents a geometric input-output analysis of hidden modes in distance-based formation control. We study the linearized dynamics under a gradient control law to characterize the system's structural limitations and their dynamic consequences. Our main contribution is a unified geometric framework for uncontrollable modes. We first prove that uncontrollable rigid-body modes are exactly characterized by a global rotational subspace, mathcal{R}_i. To generalize this, we introduce the local rotational subspace, mathcal{T}_i, which establishes a strict geometric bound on the hidden deformations for minimally connected actuators (i.e., when the actuator node has as many neighbors as the dimension of the space). Finally, we demonstrate the dynamic implications of this structure by proving that the system's ability to recover its shape is determined by an input's alignment with the local component of the standard rotational rigid-body mode, directly linking the geometry of hidden modes to disturbance rejection. We illustrate our results with a case study.

Keywords: Control of networks, Distributed control and estimation, Multi-agent systems
Abstract: Bearing-only formation tracking control scheme based on control barrier functions (CBFs) for nonholonomic agents with static obstacle avoidance is investigated in this work. A relative-degree one CBFs is constructed for safety guarantee on the motion of control points, ahead of the centers, of the agents in a bearing-only formation maneuvering. The safety condition is then enforced as inequality constraints in a quadratic programming (QP) formulation that modifies the nominal formation control flows of the agents in a minimal way to avoid the obstacles. The optimal solution to the QP can be computed in closed-form and has an intuitive geometrical interpretation. Under the formation tracking control scheme, the formation system is shown to be uniformly stable and converges to the target formation asymptotically while obstacle avoidance is ensured.

Keywords: Multi-agent systems, Control of networks
Abstract: Achieving desired formations is a fundamental and important task for multiagent systems in engineering practice. This task generally involves two sequential stages: formation shape design and formation controller design. In most existing studies, the desired formation is predetermined and assumed to be graphically rigid, which inherently implies a centralized formation shape design. Therefore, it is still an important but challenging problem for each agent to check in a distributed manner whether it forms a rigid formation with its neighbors. To address this challenge, we propose an interesting concept, rigid formability, to check those subsets forming rigid sub-formations in a large-scale multiagent system. As a starting point, this work focuses on checking the rigid formability of angle-constrained multiagent systems in a distributed manner. Numerical simulations are performed to confirm the validity of the theoretical results.

Keywords: Multi-agent systems, Distributed control and estimation, Statistical analysis
Abstract: When localizing a target through two cooperative agents using distance and angle measurements, although the relative position of the target with respect to the agents can be directly obtained through their triangular geometry, the localization error may vary significantly when measurement noise exists. Motivated by this, this paper aims to investigate the sensitivity of the target localization error when the target and agents are in different geometric configurations. Firstly, this paper presents a geometric analysis, revealing the impact of deviations in various measurement quantities on the localization error of the target. Secondly, sensitivity analysis tools are utilized to conduct global and local sensitivity analyses on the relative position of the target with respect to the agents. The results of the sensitivity analysis indicate that the target’s relative position is less sensitive to the measurement noise of quantities with lower sensitivity indices. Furthermore, the differential-based sensitivity analysis provides a foundation for minimizing the localization error. Specifically, the localization error bound depends on the geometric configuration of the two agents and the target, as well as noise parameters. Finally, simulation results validate the effectiveness of the proposed method.

Keywords: Multi-agent systems, Control of networks, Control under communication constraints
Abstract: This work develops a symmetry-based framework for formation control on cycle graphs using Dihedral point-group constraints. We show that enforcing inter-agent reflection symmetries, together with anchoring a single designated agent to its prescribed mirror axis, is sufficient to realize every C_{nv}-symmetric configuration using only n-1 communication links. The resulting control laws have a matrix-weighted Laplacian structure and guarantee exponential convergence to the desired symmetric configuration. Furthermore, we extend the method to enable coordinated maneuvers along a time-varying reference trajectory. Simulation results are provided to support the theoretical analysis.

Keywords: Data-driven control theory, Learning methods for control
Abstract: This paper focuses on the analysis of the Region of Attraction (RoA) for unknown autonomous dynamical systems. A data-driven approach based on the moment-Sum-of-Squares hierarchy is proposed, enabling novel RoA outer approximations despite reduced information on the dynamics. The main contribution consists of bypassing the system model and, hence, the recurring constraint on its polynomial structure. Numerical experiments showcase the influence of data on learned approximating sets, highlighting the potential of this method.

Keywords: Extremum seeking and model free adaptive control, Nonlinear adaptive control
Abstract: In this paper, we introduce a model-free, real-time, dynamic optimization and control method for a class of rigid body dynamics. Our method is based on a recent extremum seeking control for vibrational stabilization (ESC-VS) approach that is applicable to a class of second-order mechanical systems. The new ESC-VS method is able to stabilize a rigid body dynamic system about the optimal state of an objective function that can be unknown expression-wise, but assessable through measurements; the ESC-VS is operable by using only one perturbation/vibrational signal. We demonstrate the effectiveness and the applicability of our ESC-VS approach via three rigid-body systems: (1) satellite attitude dynamics, (2) quadcopter attitude dynamics, and (3) acceleration-controlled unicycle dynamics. The results illustrate the ability of our ESC-VS to operate successfully as a new methodology of optimization and control for rigid body dynamics.

Keywords: Extremum seeking and model free adaptive control, Nonlinear adaptive control
Abstract: This paper introduces a novel model-free, real-time unicycle based source seeking design. This design autonomously steers the unicycle dynamic system towards the extremum point of an objective function or physical/scalar signal that is unknown expression-wise, but accessible via measurements. A key contribution of this paper is that the introduced design converges exponentially to the extremum point of objective functions (or scalar signals) that behave locally like a higher-degree power function (e.g., fourth-degree polynomial function) as opposed to locally quadratic objective functions, the usual case in literature. We provide theoretical results and design characterization, supported by a variety of simulation results that demonstrate the robustness of the proposed design, including cases with different initial conditions and measurement delays/noise. Also, for the first time in the literature, we provide experimental robotic results that demonstrate the effectiveness of the proposed design and its exponential convergence ability. These experimental results confirm that the proposed exponentially convergent extremum seeking design can be practically realized on a physical robotic platform under real-world sensing and actuation constraints.

Keywords: Iterative and repetitive learning control
Abstract: In linear time-invariant systems, the sensitivity function to disturbances is designed under a sensitivity tradeoff known as the waterbed effect. To compensate for a quasiperiodic disturbance, a quasiperiodic disturbance observer using time delays was proposed. Its sensitivity function avoids the sensitivity tradeoff, achieving wideband harmonic suppression without amplifying aperiodic disturbances or shifting harmonic suppression frequencies. However, its open-loop transfer function is not rational and does not satisfy the assumptions of existing Bode sensitivity integrals due to its time delays. This paper provides Bode-like sensitivity integrals for the quasiperiodic disturbance observer in both continuous-time and discrete-time representations and clarifies the avoided sensitivity tradeoff with time delays.

Keywords: Learning methods for control, Data-driven control theory
Abstract: Model Predictive Control (MPC) relies on dynamical models to compute optimal control actions in a receding-horizon fashion. When accurate models are unavailable or expensive to obtain, data-driven predictive control (DDPC) offers a viable alternative for optimal control synthesis. While the linear case is now relatively well understood, both the theoretical foundations and practical implementations for nonlinear systems remain less developed. In this paper, we proposed an effective approach toward closing this gap by combining locally linear data-driven approximations that, when iteratively refined, yield a sequential data-driven quadratic programming algorithm for nonlinear DDPC. The proposed DDPC framework is validated through numerical simulations on an inverted pendulum benchmark.

Keywords: Time/parameter varying system identification, Model reference adaptive control
Abstract: Model Reference Adaptive Control Systems (MRACS) offer excellent responsiveness but often fail to achieve parameter convergence. While conventional methods inject probing signals to ensure convergence, they typically degrade control performance. This study proposes a control error-based auto-elimination scheme that adaptively regulates the probing signal without predefined timing. This approach enables parameter identification during transient phases and automatically suppresses the signal once tracking is achieved. Furthermore, it allows for signal re-injection upon detecting performance degradation, realizing a compatible balance between identification and control. Simulations under varying plant conditions confirm the effectiveness of the proposed structure.

Keywords: Linear system identification, Control of networks
Abstract: In this paper we propose a novel adaptive online optimization algorithm tailored to the management of microgrids with high renewable energy penetration, which can be formulated as a constrained, online optimization problem. The proposed algorithm is characterized by a control-based design that applies the internal model principle, and a system identification routine tasked with identifying such internal model. In addition, in order to ensure the constraints are verified, we integrate a projection onto the constraint set. We showcase promising numerical results for the microgrid use case, highlighting in particular the enhanced adaptability of the proposed algorithm to changes in the internal model. The performance of the proposed algorithm is shown to outperform state-of-the-art alternative in the long-term, ensuring efficient management of the grid.

Keywords: Linear system identification
Abstract: Following an external perturbation, humans recover balance through a coordinated response of the ankle and hip joints, but the underlying joint-torque strategy that produces this motion cannot be measured non-invasively. The objective of this study is to quantify participant specific ankle–hip torque response strategies under small impulsive backward perturbations. Using measured ankle and hip joint angles and the perturbation force, ankle, and hip joint torque trajectories were estimated by applying direct collocation to a sagittal-plane double inverted pendulum model. Functional principal component analysis (FPCA) was then applied to the reconstructed torque trajectories to extract dominant temporal torque patterns and to summarize between-participant differences in a low-dimensional score space. As a quantitative baseline, standard inverse dynamics computed from the measured kinematics with butterworth filtering was implemented and compared with direct collocation using open-loop forward-replay error; the dynamics residual on the measured trajectory is reported in the appendix. For nine healthy participants, the first two principal components explained most of the between-participant variability and corresponded to the initial response magnitude and the late-phase residual regulation. Together with the baseline comparison, the FPCA scores provide a compact descriptor of individual balance and a basis for future clustering and personalized design of wearable balance-assistive devices.

Keywords: Linear system identification, Data-driven control theory
Abstract: In this paper, we propose a novel online optimization algorithm built by combining ideas from control theory and system identification. The foundation of our algorithm is a control-based design that makes use of the internal model of the online problem. Since such prior knowledge of this internal model might not be available in practice, we incorporate an identification routine that learns this model on the fly. The algorithm is designed starting from quadratic online problems but can be applied to general problems. For quadratic cases, we characterize the asymptotic convergence to the optimal solution trajectory. We compare the proposed algorithm with existing approaches, and demonstrate how the identification routine ensures its adaptability to changes in the underlying internal model. Numerical results also indicate strong performance beyond the quadratic setting.

Keywords: Linear system identification, Data-driven control theory
Abstract: The identifiability and solution of the inverse optimal control (IOC) for linear quadratic regulators (LQR) has been widely investigated. Most of the existing works rely on demonstrations under optimal control, whereas suboptimal ones have rarely been considered. In this paper, we study the IOC problem for discrete-time finite-horizon (DTFH) LQR under the optimal control corrupted by an additive noise. We prove that the IOC problem remains well-posed through establishing the corresponding model identifiability. An approximate solution is obtained by a single-level statistically consistent estimator using second-moment to minimize the dynamic mismatch. The second-moment approach reduces the computational complexity of the residual function compared with the commonly used risk function approach. This estimator does not require access to the input sequence or prior knowledge of the noise covariance. Its effectiveness is verified through numerical examples.

Keywords: Linear system identification
Abstract: Traditional system identification with multisine inputs relies on uniform sampling and periodic excitation to preserve Fourier orthogonality and avoid spectral leakage, limiting its use in scenarios with irregular sampling or nonperiodic inputs. This work investigates continuous-time system identification under nonperiodic multisine excitation and nonuniform sampling. We develop a nonparametric frequency response function estimator suited to such conditions and design irregular sampling schemes that enhance the informativeness of measurements and reduce spectral leakage. The proposed sampling scheme improve the statistical accuracy of system identification in settings where periodic excitation is impractical.

Keywords: Linear system identification, Data-driven control theory
Abstract: Accurate spindle modeling is essential for the control and optimization of power consumption in machining processes. This paper addresses closed-loop continuous-time system identification for spindle speed control in machine-tool applications featuring a parallel PI controller and a reference prefilter. Two algorithms are proposed: a simplified instrumental-variable method assuming white measurement noise, and an extended version that accounts for colored noise through optimal whitening. Monte Carlo simulations on a benchmark process demonstrate that the colored-noise method yields more accurate and consistent parameter estimates, particularly under low signal-to-noise ratios.

Keywords: Model predictive control of hybrid systems, Reachability analysis, verification and abstraction of hybrid systems, Stochastic hybrid systems
Abstract: This work presents a safe motion planning framework that addresses the challenge of ensuring probabilistic safety guarantees when the behavior of uncontrollable agents depends on both their own state and the state of the controllable system. We model the uncontrollable agent dynamics as random variables drawn from unknown state-dependent distributions, which are learned from offline data using neural networks. To provide probabilistic guarantees on pre- diction errors, we employ split conformal prediction to construct region-specific, time-dependent uncertainty bounds. These uncertainty bounds are integrated into a model predictive control formulation, resulting in a chance-constrained optimal control problem that ensures safety with high probability over the planning horizon. To solve this non-convex, discontinuous optimization problem, we propose a two-loop iterative sequential convex programming algorithm. The inner loop solves convexified subproblems with fixed error bounds, while the outer loop refines these bounds based on updated control sequences. We establish convergence guarantees under mild regularity conditions and demonstrate that the algorithm returns solutions satisfying the KKT optimality conditions. The framework is validated in a high-fidelity autonomous driving simula- tor with interactive pedestrians. Experimental results demonstrate that our approach achieves superior safety and efficiency compared to baseline methods, with success rates exceeding 99.5 percent while maintaining higher average speeds in multi-pedestrian scenarios.

Keywords: Multi-agent systems, Hybrid and switched systems modeling, Consensus
Abstract: This paper proposes a novel framework for multi-agent systems that ensures seamless switching among sequential models while guaranteeing exact, prescribed-time convergence. By employing continuous, time-varying weighting functions within a composite control law, the strategy interpolates between regimes to maintain command input continuity. The synchronization logic utilizes LaSalle’s invariance principle via hybrid time-domain transformations, avoiding discontinuous feedback and Lyapunov decay inequalities. Theoretical analysis and numerical simulations confirm that integrating bio-inspired potentials with temporal compression achieves smooth, scalable, and finite-horizon coordination under frequent model transitions.

Keywords: Reachability analysis, verification and abstraction of hybrid systems, Fault detection and diagnosis
Abstract: This paper proposes a fault detection strategy for singular system subject to unknown disturbance. Based on the neural network and unknown input observer, the predicted disturbance is partially decoupled, and the rest is attenuated by solving the optimization problem. According to the set-membership technique, outer approximation zonotope for the irregular intersection between iterative update set and measured strip of neural network output weight is calculated to facilitate the generation of the desired threshold, and further becomes more tight by Frobenius norm minimization technique. A numerical example is provided to testify the effectiveness.

Keywords: Reachability analysis, verification and abstraction of hybrid systems, Markov decision process, Stability and stabilization of hybrid systems
Abstract: In this paper, we present a computational approach to certify almost sure reachability for discrete-time polynomial stochastic systems by turning drift–variant criteria into sum-of-squares (SOS) programs solved with standard semidefinite solvers. Specifically, we provide an SOS method based on two complementary certificates: (i) a drift certificate that enforces a radially unbounded function to be non-increasing in expectation outside a compact set of states; and (ii) a variant certificate that guarantees a one-step decrease with positive probability and ensures the target contains its nonpositive sublevel set. We transform these conditions to SOS constraints. For the variant condition, we enforce a robust decrease over a parameterized disturbance ball with nonzero probability and encode the constraints via an S-procedure with polynomial multipliers. The resulting bilinearities are handled by an alternating scheme that alternates between optimizing multipliers and updating the variant and radius until a positive slack is obtained. Two case studies illustrate the workflow and certifies almost-sure reachability.

Keywords: Stability and stabilization of hybrid systems, Hybrid and switched systems modeling
Abstract: We address the problem of stability and ultimate boundedness of nominally linear-affine switched systems with uncertain Lipschitz nonlinearities under dwell-time constraints. In particular, we propose a sampled-data switching law based on a state observer and on Lyapunov-Metzler inequalities, accounting for the sampled-data output measurements. We derive time-dependent LMI conditions for global asymptotic stability – or, in the presence of switching affine terms, ultimate boundedness – of the closed-loop system, and we provide an estimate of the average quadratic cost and a bound on its maximum deviation from the actual cost. We also discuss the feasibility of the derived LMIs; in particular, we show how to incorporate the observer gains into the matrix inequalities, provide equivalent reduced-order LMI conditions, and prove that the LMIs can be made time-independent through discretisation on a finite grid. Numerical examples illustrate our theoretical results and their efficacy.

Keywords: Multi-agent systems, Consensus, Distributed control and estimation
Abstract: This paper presents a constructive framework for synthesizing provably safe controllers for nonlinear multi-agent systems subject to bounded disturbances. The methodology applies to systems representable in Brunovsky canonical form, accommodating arbitrary-order dynamics in multi-dimensional spaces. The central contribution is constraint tightening method that formally couples robust error feedback with nominal trajectory planning. The key insight is that the design of an ancillary feedback law, which confines state errors to a robust positively invariant (RPI) tube, simultaneously provides the information needed to ensure the safety of the nominal plan. The geometry of the resulting RPI tube is leveraged via its support function to derive state-dependent safety margins. These margins systematically tighten the high relative-degree exponential control barrier function (eCBF) constraints on the nominal planner, guaranteeing that any nominal trajectory satisfying the tightened constraints yields a provably safe trajectory for the true, disturbed system. The planner is implemented within a distributed Model Predictive Control (MPC) scheme that optimizes performance while inheriting the robust safety guarantees.

Keywords: Control engineering curricula
Abstract: The position regulation of a mass sliding on a rectilinear guide is addressed by two feedforward-feedback control strategies: a set-point filtered two-degree-of-freedom (2-DOF) scheme and a plant inversion scheme. The design of these control systems complies with an amplitude constraint on the force applied to the mass and the requirement of no overshoot or oscillations in the mass motion. A numerical example compares the performances of the two schemes, evidencing the advantages of the plant inversion control. The exercise may be suitable for teaching in an undergraduate first course in automatic control.

Keywords: Control engineering curricula, Repositories for control education
Abstract: Closed loop control systems are based on negative feedback. The output of the plant to be controlled is measured, and this value is compared to the reference signal. The obtained error signal provides the input signal to a controller which creates the actuating signal to the input of the plant. The output signal should reach the required value or follow the shape of the reference signal after deceasing of the transients. The control system should track the reference signal and also eliminate the effect of the disturbances and of the uncertainties. Stability is a basic issue in feedback control systems. This means that the transients should decrease and the steady state should be reached. The dead time in the process, delayed action strongly influences stability. Ensuring stability in case of controlling an oscillating or an unstable process (e.g. inverted pendulum) is also a challenge. The intended learning outcome is understanding the problem of stability, illustrating the problem analysing the proportional control of a pure dead time process and the ability to use different methods learned in the theoretical material to evaluate stability. Stability can be analysed based on the location of the poles in the closed loop system, or in the frequency domain using the Nyquist stability criterion. Several examples are given to check stability conditions of a continuous control system using different methods. The aim of the exercises is to give a deeper understanding of stability and expertise in using stability investigation methods.

Keywords: Repositories for control education
Abstract: This paper presents an exercise constituted by a sequence of open-ended, design-and-explain questions that guide students through the theoretical design and practical simulation of an Active Disturbance Rejection Control (ADRC) for a pendulum. The exercise is structured as a theoretical ``paper-and-pencil'' part followed by a practical part performed in MATLAB/Simulink. The exercise requires students to leverage preexisting knowledge of system modeling, state-space representation, and observer design. The resource is intended for undergraduate or master's students in control engineering who have completed an introductory automatic control course, aiming to bridge the gap between theoretical concepts and practical implementation.

Keywords: Repositories for control education, Control education laboratories, Continuing control education
Abstract: This paper presents two exercises in control systems theory targeted towards learners taking a B.Sc.-level course on the topic. The first exercise is designed to serve as a motivating example of the ubiquity of control theory and its usefulness in solving problems in seemingly unrelated contexts. The famous Fibonacci sequence serves as a vehicle for analyzing homogeneous linear recurrence relations with constant coefficients. The second exercise is intended to illustrate a common misconception regarding the application of Internal Model Control principles for robust regulation, namely, achieving zero steady-state error in a Linear Time Invariant setting. Basic tools covered in an introductory course on control systems are employed throughout.

Keywords: Repositories for control education, Control education learning analytics, Control engineering curricula
Abstract: This paper presents a collection of conceptual stability questions for continuous and discrete time systems, both in open-ended and multiple choice questions formats, designed to target typical misconceptions regarding simple (a.k.a. marginal) stability and convergence properties of equilibria, and the distinction between local and global behaviors. The intended aids are pen-and-paper reasoning and schematic phase-portrait intuition. The target audience is advanced BSc or early MSc students with basic knowledge of dynamical systems and equilibrium notions.

Prerequisite LOs (PLOs): - Recall definitions of equilibrium, simple (marginal) stability, asymptotic stability, and convergence. - Manipulate and reason about nonlinear continuous- and discrete-time dynamical systems. - Interpret trajectories and qualitative behavior using phase-portrait arguments.

Intended LOs (ILOs): - Distinguish convergence from simple stability in nonlinear systems. - Diagnose violations of the delta--varepsilon based definition of simple stability. - Understand how switching, saturation, or history-dependent dynamics may affect equilibrium properties.

Keywords: Repositories for control education, Internet based control education
Abstract: This paper presents three exam questions in control design, selected from many years of teaching both undergraduate and graduate control engineering courses. The first question demonstrates a numerical approach in which controller zeros and residue poles are placed outside the dominant pole region to meet time-domain performance specifications. Although the idea is numerical, the question is intentionally structured so that it can be solved algebraically with pen and paper. The second question illustrates the use of the Separation Principle in designing a Luenberger observer together with state feedback, highlighting the complications that arise when unobserv- able modes appear in the observer design. The final question is taken from an industrial setting: a pasteurization process modeled as a time-delay plant with an unknown delay. The problem focuses on applying Ziegler–Nichols tuning when only limited information about the plant is available.

Keywords: Decision making under uncertainty, Smart city control and optimization, AI for smart cities
Abstract: This paper proposes a fully dynamic Deep Reinforcement Learning (DRL) method for rebalancing dockless bike-sharing systems, overcoming the limitations of periodic, system-wide interventions. We model the service through a graph-based simulator and cast rebalancing as a Markov decision process. A DRL agent routes a single truck in real time, executing localized pick-up, drop-off, and charging actions guided by spatiotemporal criticality scores. Experiments on real-world data show significant reductions in availability failures with a minimal fleet size, while limiting spatial inequality and mobility deserts. Our approach demonstrates the value of learning-based rebalancing for efficient and reliable shared micromobility.

Keywords: Data centers and cloud computing, AI for smart cities, Decision making under uncertainty
Abstract: This paper investigates the coordination between power systems and data centers in uncertain environments. First, the interaction is formulated as a Stackelberg game where the retailer operator from the power systems sets dynamic electricity prices and the data center operator responds through workload scheduling and battery operations. A realistic data center model is developed to capture task-level temporal flexibility and to enable workload shifting subject to deadline and capacity constraints. Second, a reinforcement learning–based solution framework is developed in which two agents are trained for the leader and the follower. The follower agent is pre-trained offline to learn the optimal demand response under general pricing policies, and the learned follower is subsequently used to train the leader agent to derive an optimal pricing strategy. Finally, numerical experiments demonstrate that the proposed method is shown to achieve the highest renewable energy utilization while maintaining competitive net revenue compared with time-of-use (TOU) and rule-based pricing policies. Overall, the results indicate that the learning-based Stackelberg game framework effectively leverages renewable generation patterns and data center flexibility to enhance system-level performance.

Keywords: Smart city control and optimization, Decision making under uncertainty, AI for smart cities
Abstract: Urban traffic congestion presents a significant challenge for modern cities, which impacts mobility and sustainability. Traditional traffic light control systems often fail to adapt to dynamic conditions, leading to inefficiencies. This paper proposes a novel deep reinforcement learning agent for traffic light control that addresses this limitation by explicitly integrating fairness considerations for both vehicular and pedestrian traffic. Unlike prior work, our approach dynamically balances these flows based on real-time demand, moving beyond systems focused solely on vehicles. Experimental results demonstrate that our agent effectively reduces congestion while ensuring equitable service for both the categories of road users. This research contributes to a practical and adaptable solution for intelligent traffic management within the framework of smart cities, paving the way for more efficient and inclusive urban mobility.

Keywords: Smart city control and optimization, Cyber-physical urban systems, Urban energy distribution systems
Abstract: As a crucial component of energy supply in smart cities, microgrid systems require accurate state estimation to support efficient decision-making and control in smart city energy systems. Although deploying numerous sensors enhances estimation reliability, it also increases the system's susceptibility to unreliable measurements caused by unknown disturbances and random link failures in communication networks. To address these challenges, this paper investigates the distributed state estimation problem for microgrid systems subject to both measurement outliers and random communication link failures. The proposed distributed state estimation algorithm comprises two key components: first, a variational Bayesian-based local state estimator that adaptively mitigates external disturbances; second, a robust consensus-based fusion mechanism that effectively integrates local estimates from neighboring sensors while maintaining estimation performance under random link failures.

Keywords: Smart city control and optimization, Urban energy distribution systems, Cyber-physical urban systems
Abstract: This paper investigates a passivity-based hierarchical frequency control framework for frequency regulation and system-wide stability in heterogeneous nonlinear power systems. The power network consists of synchronous generators (SGs) with turbine/governor dynamics, virtual synchronous generators (VSGs), and droop-controlled inverters (DCIs) interconnected through nonlinear power flows. Since such systems generally lack strict passivity due to heterogeneous dynamics and relative degrees, we employ the passivity-short concept to capture their input–output behavior in a unified manner. Based on this framework, we design a hierarchical control strategy comprising local controllers to stabilize the overall system and a distributed control law using limited communication for global cooperation. The proposed method guarantees overall system stability and enhances the frequency regulation, as demonstrated by simulation results.

Keywords: Cyber-physical urban systems, Human-centric automation/AI Systems, and human agency, Decision making under uncertainty
Abstract: Emergency medical services (EMS) are part of the critical infrastructure of cyber-physical urban systems (CPUS), and mass casualty incidents (MCIs) place particular strain on prehospital triage. This paper introduces TARIS (Triage Assistance using Real-time Intelligence & Support), a conceptual assistance system that combines artificial intelligence (AI) and augmented reality (AR) for triage support in such events. Following a design science process, TARIS links wearable sensors, AR-based guidance and AI-generated triage suggestions in a modular cloud–fog architecture spanning physical and cyber layers. Qualitative expert interviews suggest that practitioners see potential to reduce workload, support decision consistency and improve information flow along the rescue chain.

Keywords: Model predictive control
Abstract: Model Predictive Control (MPC) is widely used for constrained setpoint tracking problems. Due to lack of anticipation of future setpoint changes, tracking MPC often suffers from transient performance degradation or loss of feasibility under significant setpoint variations. This paper proposes a reference-preview MPC that improves transient tracking performance while guaranteeing persistent feasibility under arbitrary setpoint variations. The controller explicitly incorporates a finite-horizon preview of future setpoint values into the optimization problem, and introduces a time-varying artificial reference sequence as additional decision variables. A maximal tracking admissible invariant set is constructed from a parameterization of all feasible steady states and then employed as the terminal constraint, eliminating the need for online terminal set recomputation. We prove recursive feasibility and asymptotic stability of the closedloop system. A simulation example demonstrates that the proposed approach achieves faster, anticipatory responses compared with existing tracking MPC methods that do not utilize future setpoint knowledge.

Keywords: Model predictive control
Abstract: Model Predictive Control (MPC) for tracking provides a powerful framework for tracking potentially unreachable and time-varying references while ensuring recursive feasibility, stability and convergence to the best reachable reference through the introduction of an artificial reference. In this work, we extend existing results in two ways. First, we address not only setpoint tracking but tracking of periodic references and make use of an formulation that does not require terminal ingredients, thereby reducing design effort. Second, we provide rigorous bounds on the closed-loop performance and show that for suitable parameters the MPC for tracking input becomes optimal as the prediction horizon tends to infinity.

Keywords: Model predictive control, Design methods for data-based control, Linear systems
Abstract: Model Predictive Control (MPC) offers safe and near-optimal control but suffers from high computational costs. Approximate MPC (AMPC) mitigates this by learning a cheaper surrogate policy, typically by training a neural network on state-MPC input pairs. Generating training data is a major bottleneck, requiring solving the MPC for numerous states sampled from its feasible set. Since this feasible set is implicitly defined and unknown, efficient sampling is nontrivial but crucial. We propose the linear MPC Hit-and-Run (LMPC-HR) sampler for linear MPC with polyhedral constraints. We identify the feasible set boundaries along search directions, a crucial step within HR, by formulating the problem as a convex linear program, replacing expensive iterative searches with a single optimization step. A numerical study demonstrates that LMPC-HR reduces the computational cost of generating uniformly distributed samples from the feasible set by an order of magnitude compared to standard baselines.

Keywords: Model predictive control, Probabilistic robustness, Stochastic optimal control problems
Abstract: This paper considers stochastic linear time-invariant systems subject to time-averaged state-constraint violation bounds, without assuming knowledge of the disturbance distribution. We propose a disturbance-adaptive model predictive control (DAD-MPC) framework, which adjusts the confidence level and the induced disturbance set based on measured constraint violations. Using a robust invariance method, DAD-MPC ensures recursive feasibility and guarantees robust or asymptotic bounds on the average violation rate. Additionally, the bounds remain valid even with an inaccurate disturbance model, enabling the use of data-driven disturbance quantification methods such as conformal prediction. Simulation results demonstrate that the proposed approach reduces closed-loop cumulative cost compared to state-of-the-art methods across different target violation rates, while satisfying average violation bounds.

Keywords: Model predictive control, Sampled-data/digital control, Linear systems
Abstract: In adaptive-sampling control, the control frequency can be adjusted during task execution. Ensuring that these changes do not jeopardize the safety of the system being controlled requires attention. We introduce robust M-step hold model predictive control (MPC) to address this. Our formulation provides robust constraint satisfaction for an uncertain discrete-time system model subject to an adaptable multi-step input hold ("M-step hold"). We show how to ensure recursive feasibility of the MPC utilizing M-step hold extensions of robust invariance, and demonstrate how to enable safe adaptive-sampling control via the online selection of M. We evaluate the utility of the robust M-step hold MPC formulation in a cruise control example.

Keywords: Model predictive control, Uncertain systems, Data-driven robust control
Abstract: We develop a learning-based control algorithm for constrained trajectory tracking by nonlinear systems with unknown dynamics by using data obtained on the fly. The algorithm consists of three steps: first, it uses an observer-based differentiator to estimate the systems’ state derivatives; second, it uses limited information on the systems dynamics (such as Lipschitz constants in a given set) and state measurements obtained on the fly, i.e., from a single system trajectory, to compute in real-time a set-based over-approximation of the unknown dynamic terms; this over-approximation leads to a differential-inclusion-based dynamic approximation that is updated on the fly as new state measurements are obtained. Third, it uses the overapproximation in a Nonlinear Model Predictive Control procedure to achieve tracking of a predefined trajectory while complying with state and input constraints. We provide theoretical guarantees and simulations results on two underactuated systems.

Keywords: Adaptive control design, Learning methods for optimal control, Linear systems
Abstract: The study presents an online Policy Iteration (PI) algorithm for the adaptive Linear Quadratic Regulator (LQR) problem under stochastic disturbances. The proposed method recursively estimates the value function using an Instrumental Variable (IV) technique from real-time input–output data, eliminating the requirement for pre-collected data. In the subsequent policy improvement step, the state feedback gains are updated via a gradient descent law with a properly chosen step size, enabling continuous online adaptation while ensuring smooth convergence and a monotonic decrease of the value function toward the optimal value. The main theoretical contributions include the convergence of the proposed online PI algorithm and the analysis of the convergence properties of the parameter estimation. For further verification, numerical simulations are presented to demonstrate the effectiveness and practical applicability of the method without requiring full knowledge of the system models under stochastic disturbances.

Keywords: Adaptive control design, Learning methods for optimal control, Uncertain systems
Abstract: This article presents a method for adaptive control of pure feedback nonlinear plants. Control is designed based on the Actor-Critic reinforcement learning method. A modification of the Actor-Critic algorithm is proposed, which lacks the feature of zeroing control. An additional coefficient has been introduced into the tuning algorithm to stabilize the control parameters. Adaptive control is approximated by radial base functions network. The design of control is carried out in stages, based on the structure of the control object. The convergence analysis of the control system and numerical studies using the example of vessel trim control are carried out.

Keywords: Design methods for data-based control, Learning methods for optimal control, Robust controller synthesis
Abstract: Many fleet-control scenarios require consistent task performance across clients with heterogeneous dynamics and limited local data. We propose an uncertainty-aware clustered federated identification-to-control pipeline. Each client performs online exponentially weighted recursive least-squares identification during task execution and uploads parameter and covariance estimates. The server clusters clients in parameter space using a Mahalanobis distance, computes precision-weighted cluster models, and designs a two-degree-of-freedom controller per cluster. On a lateral vehicle dynamics benchmark for yaw-rate tracking during lane changes, with heterogeneity from speed and payload, the method recovers latent regimes and achieves near-nominal fleet-wide tracking. Compared with nominal, FedAvg-based, and Euclidean-clustering baselines, uncertainty-aware clustering yields more consistent tracking across regimes at modest communication and computation cost.

Keywords: Design methods for data-based control, Nonlinearity learning from data
Abstract: We present a big data-driven predictive control approach in the behavioural systems framework to control nonlinear processes in an operation region where only limited data are available. To control the nonlinear system behaviour in the operation region (target) that has a limited number of data trajectories, we utilise the large number of data trajectories in another operation region (source), under the assumption of the existence of a bijective and differentiable mapping of data trajectories between the two regions. The proposed approach consists of three steps. The first step aims to transfer the probability density function of the data trajectory from a source operation region to another target operation region. The second step is to approximate local linear sub-behaviours in the target operation region based on the transferred probability density function. The last step is to utilise the local linear sub-behaviours for online big data-driven predictive control. The proposed approach is illustrated by an example of controlling a vanadium flow battery.

Keywords: Design methods for data-based control, Nonlinearity learning from data, Data-driven robust control
Abstract: Reservoir computing is used to develop a model-free inversion-based controller for a nonlinear system. For this, an echo state network is considered as reservoir that is trained to learn the output-to-input map of the system from training data. This results in a model-free (nonlinear) tracking controller by driving the trained reservoir with the measured output and a desired reference trajectory. To address training errors and to enhance the robustness of the model-free control approach, an adaptation mechanism is proposed to adjust the output layer of the reservoir online. In addition, the local asymptotic stability of the closed-loop control system is analyzed. Finally, the control approach is validated using experimental data and is implemented to control a nonlinear three tank system. The obtained experimental results clearly confirm the performance of the model-free control concept.

Keywords: Design methods for data-based control, Stability of nonlinear systems, Nonlinearity learning from data
Abstract: Driving a system to its target within a prescribed time is essential in time-critical operations where the convergence time is a hard mission constraint. Exact convergence at the prescribed time requires unbounded gains incompatible with actuator bounds, and existing designs require analytical or restrictive structural knowledge of the dynamics. This paper proposes a data-driven prescribed-time (PT) control framework for mechanically fully actuated nonlinear systems with unknown dynamics. A deep Koopman operator lifts the system into a linear time-invariant representation, on which a bounded feedback derived from a parametric Lyapunov equation regulates the physical state, with a robust term absorbing the matched closure error of the lift, achieving practical PT convergence. The design admits a closed-form ellipsoidal domain of attraction determined by the actuator bound, initial state, and prescribed time. Numerical simulations on close-range rendezvous with an eccentric chief orbit confirm convergence within the prescribed time without violating the actuator bound.

Keywords: Stability of nonlinear systems, Interconnected nonlinear systems
Abstract: In this paper we present dissipativity analysis of clockwise Preisach hysteresis operators, which is subsequently used to establish the absolute stability of an LTI system feedback interconnected with such hysteresis element. First, we show the construction of a storage function for clockwise Preisach hysteresis operators that satisfies the clockwise dissipativity inequality, where the supply rate is given by ydot u with (y,u) be the output and input of the hysteresis operators, respectively.

Correspondingly, we show that a negative feedback interconnection of a negative imaginary system with a clockwise Preisach operator is absolutely stable.

Keywords: Saturation and discontinuity, Stability of nonlinear systems, Convex optimization
Abstract: This paper presents a global and local stability analysis of continuous-time systems with saturation or deadzone nonlinearities. The approach uses an implicit ramp-based model and Piecewise Quadratic Lyapunov Functions (PWQ LF). A key contribution is the enhancing of PWQ LF partition using fictitious ramp signals. Good assumptions about these ramps allow us to write properties that lead to less conservatism in the Linear Matrix Inequalities (LMIs), ensuring PWQ non-negativity and estimating the region of attraction using Lyapunov’s theory. Numerical results demonstrate the effectiveness of the method.

Keywords: Stability of nonlinear systems, Lyapunov methods, Interconnected nonlinear systems
Abstract: This paper pursues dissipativity-based control design for positivizing and stabilizing dynamical systems with respect to freely specified interior equilibria in positive state spaces. Recently, the use of asymmetrically scaled sectorial (ASSEC) supply rates was proposed to construct a single Lyapunov function establishing positivization and stabilization simultaneously. In contrast to symmetric dissipativity, such as Lp gain, passivity, and input-to-state stability, uncommon denominators arising from ASSEC supply rates not only deny the standard technique to justify weighted summation of the supply rates but also require cascade and feedback connections to be treated separately. This paper proposes a mechanism of flexibility propagation in Lyapunov function synthesis for cycle networks. The flexibility is useful to confirm propagation of a single extra decay across a network to form an unbounded total decay. The flexibility also unifies cascade and feedback Lyapunov function formulas, and establishes the continuity of the Lyapunov function with respect to network connection strength. Its usefulness and effectiveness are illustrated by a gradient descent controller design example.

Keywords: Nonlinear control of switched & hybrid systems, Switching stability and control, Lyapunov methods
Abstract: Over the past few decades, several techniques have been presented for the stabilization control of Switched Affine Systems (SASs), most of which are based on the solution of Linear Matrix Inequalities (LMIs), making them numerically efficient. In the case of interconnected SASs though, the performance of the numerical solvers can become severely degraded, specially in the case of large-scale systems composed of multiple SASs. In addition, some systems exhibit two distinct timescales (fast and slow) and they are usually referred to as singularly perturbed systems, for which analysis and control design suffer from poor conditioning. In this paper, we propose a constructive control design method for the stabilization of an interconnection of heterogeneous SPSASs overcoming the aforementioned difficulties. We focus on the case where the SPSASs are coupled by an LTI system that is slow compared to the fast dynamics of the SPSASs, examples of which can be found in power electronic applications. The proposed controller is decentralized and independent of the singular perturbation parameter.

Keywords: Nonlinear observers and filters, Interconnected nonlinear systems, Sliding mode control
Abstract: The problem of online parameter estimation for output-coupled nonlinear systems is discussed using exact differentiators to determine the time derivatives of the available measurements. For this purpose a class of nonlinear systems is considered that can be brought into a regressor form using the associated observability map. First, the case of a single system is addressed and sufficient conditions for the parameter estimation algorithm are established using Lyapunov's direct method. This is showcased for the Van der Pol oscillator. Then, the case of output-coupled systems is considered and a generalization of the first result is provided. This second result is then used for the parameter estimation of coupled Kuramoto oscillators, leading to improved convergence conditions in comparison to the recent study in Lundt et al. (2025). All case example are accompanied with numerical simulation results for discussion of the potential of the proposed approach.

Keywords: Interconnected nonlinear systems, Control of complex systems, Adaptive control design
Abstract: While investigating the onset of synchronization in networks of oscillatory systems, a rather intuitive assumption is to consider that all the oscillators are identical. However, there exist oscillatory systems that seem to defy this ideal scenario, in such way that the oscillators only achieve synchronization when they are heterogeneous. Here, we will exploit this counter-intuitive direction. In particular, we will consider a family of phase-amplitude oscillators, for which identical synchronization can only be observed when there is certain amount of heterogeneity in the oscillators. In our approach, this heterogeneity is introduced by adding dynamic disturbances to the nodes. Specifically, each disturbance is generated by a first-order linear system which is ad hoc designed such that the disturbance vanishes once the desired synchronous behavior has been reached. In the absence of this heterogeneity, the network of homogeneous oscillators cannot be synchronized. We consider two scenarios: the case where every node in the network is perturbed and the case where the perturbations are applied to only a few nodes. In both cases, we conduct a stability analysis of the synchronous solution and the obtained results are illustrated via numerical simulations. Ultimately, the results presented here suggest that our approach successfully induces synchronization by temporarily introducing heterogeneity in the network.

Keywords: Linear parameter-varying systems, Control barrier functions and state space constraints, Robust linear matrix inequalities
Abstract: This paper proposes the extension of the recently introduced concept of quadratic repulsiveness to polytopic linear parameter varying (LPV) systems. The novel idea of the quadratic repulsiveness is the interest in driving the state of the system out of a certain undesired region of the space, described as the super-level set of a prescribed, fixed a priori sign-indefinite quadratic function. With this goal in mind, the problem of designing stabilizing and repulsive gain-scheduled controllers is cast in terms of matrix inequalities (LMIs) that can be treated with existing solvers. The proposed approach is illustrated with the case-study of an electrical circuit.

Keywords: Linear parameter-varying systems, Control of uncertain LPV systems, Switching linear systems
Abstract: This paper is concerned with robust stability analysis and gain-scheduled state feedback synthesis for discrete-time descriptor systems whose coefficient matrices are characterized by random polytopes. Since the derived inequality conditions include random variables, a sample-based method of using them is proposed, which is illustrated with numerical examples.

Keywords: Uncertain systems, Convex optimization, Sum-of-squares
Abstract: For long the S-variable approach has proved its efficiency for addressing robustness issues with respect to matrix-valued bounded, polytopic uncertainties. The results are extended in this paper to generic convex polyhedra thus allowing robustness analysis with respect to unbounded uncertainties. Results apply to rationnaly-dependent models with respect to such polyhedral uncertainties. Moreover the S-variable approach is proved to be enhanced thanks to a new affine rows-decoupled descriptor modeling of such systems. Examples of application of these results are a reinterpretation of the KYP-lemma as well as a reformulation of the links between sum-of-squares approach and the S-variable approach.

Keywords: Uncertain systems, Robust linear matrix inequalities, Lyapunov methods
Abstract: This paper deals with the problem of covariance stabilization for a class of linear stochastic discrete-time systems in the Stochastic Model Predictive Control (SMPC) framework. The considered systems are affected by independent and identically distributed (i.i.d.) additive and parametric stochastic uncertainties (potentially unbounded), in addition to polytopic deterministic uncertainties bounding the mean of the state and input parameters. The design conditions presented in this paper are formulated as Linear Matrix Inequalities (LMIs), using the S-variable approach in order to reduce the potential conservatism. These conditions are derived using a deterministic exact characterization of the covariance dynamics, the latter involves bilinear terms in the control gain. A technique to linearize such dynamics is presented, it results in a descriptor representation allowing to derive sufficient conditions for the design of a covariance-stabilizing controller. The derived condition is first compared with a known necessary and sufficient stability condition for systems without deterministic uncertainties and additive stochastic noise. Although more conservative, the proposed condition is more numerically tractable, with an LMI size scaling as O(n^2) instead of O(n^3). Then, the same condition is used to design controllers that are robust to both deterministic and stochastic uncertainties. Several numerical examples are presented for comparison and illustration.

Keywords: Lyapunov methods, Stability of nonlinear systems, Robust linear matrix inequalities
Abstract: This paper proposes a homogeneous rational Lyapunov function for exponential stability analysis of continuous-time Takagi-Sugeno fuzzy systems. The numerator and denominator are homogeneous polynomials with independent degrees, generalizing homogeneous polynomial Lyapunov functions and providing additional flexibility. Stability conditions are derived in terms of matrix inequalities and solved through an LMI-based iterative algorithm that jointly optimizes the numerator, denominator, and exponential decay rate. Unlike related rational formulations, both numerator and denominator matrices remain decision variables. Numerical examples show that the proposed approach reduces conservatism and improves certified decay rates compared with existing conditions based on polynomial Lyapunov matrices.

Keywords: Control of uncertain LPV systems, Robust controller synthesis, Robust linear matrix inequalities
Abstract: This paper proposes an iterative linear matrix inequality (LMI) based approach to compute, directly as variables of the problem, output feedback control gains for discrete-time linear parameter varying systems and the corresponding polyhedral Lyapunov function with a prescribed level of complexity. For that, the (necessary and sufficient) conditions from the literature are manipulated through the double application of Finsler's lemma, generating a new equivalent formulation where the control gains and other meaningful decision variables appear explicitly in the conditions. These new conditions are solved through a locally convergent LMI-based iterative algorithm. Numerical examples illustrate the results.

Keywords: Lagrangian and Hamiltonian systems, Disturbance rejection and input-to-state stability, Output regulation and tracking
Abstract: This paper presents a robust control framework for Pendubot swing-up and stabilization. To avoid computationally expensive optimization, a reference trajectory is generated via the time-reversal of a SD-controlled swing-down motion. Since physical friction breaks the dynamic symmetry required by this approach, a time-delay control law is designed to achieve robust finite-horizon tracking of the active joint. By introducing a virtual inertia, the controller estimates and compensates for the actuated-channel lumped disturbance, including active-joint friction, coupling effects, and model mismatch entering the first joint dynamics, without explicit regression models. Input-to-state stability of the active-joint tracking error dynamics is analytically established. Finally, a switched linear quadratic regulator provides local stabilization near the upright equilibrium. Numerical simulations validate the method's robustness against active-joint friction and a representative inertial parameter perturbation.

Keywords: Application of nonlinear analysis and design, Controller constraints and structure
Abstract: This paper addresses the attitude tracking control problem for a rigid body with attitude measurements only, when angular velocity measurements are not available. To circumvent the complexity and ambiguity associated with alternative attitude representations like Euler angles or quaternions, the attitude dynamics and the proposed control system are globally represented on special orthogonal groups. Through the design of an auxiliary variable, an efficient and practical angular velocity-free control strategy with a simple structure is proposed, which achieves asymptotic tracking of an attitude command without requiring explicit knowledge of angular velocity signals.

Keywords: Output feedback nonlinear control, Application of nonlinear analysis and design, Output regulation and tracking
Abstract: To support the precise and repeatable evaluation of fuel cell systems, testbed gas conditioning units must accurately reproduce dynamic conditions with minimal interference between flow, pressure, and temperature. This work addresses the intrinsic coupling of these variables by introducing a model- and flatness-based feedback control scheme for the gas conditioning subsystem of a commercial fuel cell test bed. The approach enables purposeful actuation of balance-of-plant components to generate desired gas conditions while minimizing cross-coupling in the remaining states. The resulting decoupled behavior is demonstrated in simulations, highlighting the method’s potential to enhance diagnostic capability and accelerate experimental development of fuel cell systems.

Keywords: Output regulation and tracking, Disturbance rejection and input-to-state stability, Application of nonlinear analysis and design
Abstract: This paper revisits the input disturbance rejection problem for a class of strict-feedforward nonlinear systems in the presence of uncertain exosystems. To address the challenges arising from exosystem uncertainties, a nonlinear internal model is introduced and incorporated into the regulator design. Unlike most existing approaches, which typically assume known exosystems and employ linear internal models, the proposed method explicitly accounts for parameter uncertainties through the construction of an augmented system embedding the nonlinear internal model. Based on this augmented formulation, a robust stabilizing controller is systematically developed. It thus provides a solution for rejecting disturbances generated by uncertain exosystems in strict-feedforward nonlinear systems using bounded controls.

Keywords: Output feedback nonlinear control, Nonlinear observers and filters, Application of nonlinear analysis and design
Abstract: The coordinated path following problem for a quadrotor UAV is addressed. Unlike trajectory tracking, the coordinated approach decouples geometric convergence to a spatial curve from the timing law governing the speed along the path. The control synthesis relies on dynamic extension to achieve uniform relative degree, followed by transformation to a cascade normal form. An extended state observer reconstructs the unmeasured velocity components and estimates lumped disturbances from position measurements only. The resulting output feedback law ensures semi-global asymptotic stability with bounded tracking error. Simulation results for straight-line and helical paths con rm the theoretical predictions.

Keywords: Output regulation and tracking, Control barrier functions and state space constraints, Disturbance rejection and input-to-state stability
Abstract: A multi-block tracking differentiator for smoothing support trajectories, considering specified constraints on dynamic features, is designed via the block control principle. This is a canonical autonomous system with corrective actions in the form of nested sigmoids. A system of double inequalities has been formalized for tuning the constant correction gains, at which the differentiator output tracks the support non-smooth trajectory with some accuracy and generates a reference trajectory, while the remaining variables generate smooth bounded derivatives of the corresponding orders. Mechanisms for varying the gains have been developed to improve the accuracy of the support trajectory approximation without losing smoothness and violating the specified constraints. The results of the numerical simulation are presented.

Keywords: Manufacturing prognostics and health management, Industrial artificial intelligence
Abstract: This paper presents Fed-LoRA, a lightweight federated learning framework for real-time monitoring in IoT-enabled connected manufacturing. The system integrates an edge–cloud architecture with event-driven model updates and external LoRA adapters, enabling safe, traceable, and resource-efficient adaptation. Edge devices perform continuous Health indicator (HI) construction, prognostics, and closed-loop control, supported by an MQTT pipeline with low latency and reliable streaming across QoS levels. Updates are triggered only when an HI-conflict detector flags inconsistent degradation trends, sending privacy-preserving summaries to the cloud. A few-sample, few-epoch federated update then trains only LoRA adapters while keeping the base model frozen, reducing computation and communication by over 90%. A validation-gated aggregation releases updates only when accuracy improves. Experiments on the Tennessee Eastman Process show 70–80% reductions in RUL prediction errors, more coherent HI behavior, and elimination of premature triggers, while requiring just 0.63% of the sample exposure of a 200-epoch centralized baseline.

Keywords: Manufacturing prognostics and health management, Industrial artificial intelligence
Abstract: In recent years, transfer learning has been widely adopted in bearing fault diagnosis owing to its ability to transfer knowledge across different operating conditions. However, under significant variations in operating conditions, such as cross-load scenarios, the distribution of fault samples in the target domain often undergoes substantial shifts. As a result, models may achieve acceptable overall accuracy while still struggling to reliably discriminate critical fault categories. To address this issue, a probability-calibrated domain-adversarial neural network (PC-DANN) is proposed in this paper for cross-load bearing fault diagnosis. When the target domain lacks labels, labeled samples from the source domain are exploited to jointly drive feature extraction and adversarial domain alignment, enabling PC-DANN to learn transferable fault features. Simultaneously, a probability calibration module is introduced at the classifier output to perform class-wise temperature scaling and bias correction on the class logits, alleviating prediction overconfidence and class-dependent logit bias under cross-load transfer. With this design, diagnostic accuracy and probability calibration in the target domain under diverse operating conditions are simultaneously improved. Cross-load unsupervised transfer experiments are conducted on a publicly available bearing dataset, and the results demonstrate that PC-DANN outperforms competing methods across multiple evaluation metrics, thereby validating its effectiveness and advantages in scenarios involving cross-load conditions and variations in the fault-class distribution.

Keywords: Manufacturing prognostics and health management
Abstract: The prediction of remaining useful life (RUL) is the key to prognostics and health management. Regarding the issue of incomplete representation of the health status in RUL prediction, an intelligent two-stage prediction method is proposed. In the first stage, health indicator (HI) cluster is constructed to represent the health status considering the changing trend of the extracted feature data. Informer is trained to obtain the HI model. In the second stage, the identified degradation points, derived through spectral clustering, are applied to a gated recurrent unit (GRU) based model for predicting the RUL. The results show that the proposed method achieves high accuracy, outperforming some traditional methods represented by convolutional neural network (CNN) and long short-term memory (LSTM) methods.

Keywords: Manufacturing prognostics and health management, Industrial artificial intelligence
Abstract: Most industrial anomaly detectors rely on a single view, focusing solely on temporal sequences or variable structures, which limits their ability to capture comprehensive dependencies. In this paper, we propose a dual-view framework jointly modeling temporal sequence and variable-relation graph, so dependencies unfolding in time and across variables are both represented. Contrastive learning in each view pulls together semantically consistent instances while pushing apart others, shaping discriminative representations without labels. Moreover, A cross-view alignment loss enforces semantic consistency between two views, enabling coherent fusion for scoring. Experiments on the Tennessee Eastman process demonstrate consistent improvements over a set of methods.

Keywords: Manufacturing prognostics and health management, Data-driven and AI-based modelling of production and logistics, Industrial artificial intelligence
Abstract: Prognostics and health management has proven effective for predicting failures and optimising maintenance in tangible assets. However, the growing reliance on software-driven, cloud-based and interconnected infrastructures extends reliability challenges to intangible assets such as applications, configurations and data flows. These assets degrade through mechanisms that differ fundamentally from physical deterioration and are not fully addressed in traditional prognostic approaches. To clarify how predictive maintenance is studied across this heterogeneous landscape, this paper conducts a Systematic Mapping Study covering work on intangible assets and hybrid systems that integrate tangible and intangible assets interactions. The results show that while prognostic methods for tangible assets remain a crucial foundation, the literature provides limited coverage of degradation processes affecting intangible assets and offers almost no approaches for predictive maintenance in hybrid systems where tangible and intangible components interact and mutually influence each other’s degradation. Existing contributions also remain fragmented across disciplinary boundaries, preventing the emergence of a coherent cross-domain understanding. Based on the mapping, the paper discusses a cross-layer perspective intended to support future development of a unified framework for predictive maintenance applicable to tangible, intangible and hybrid assets.

Keywords: AI and learning-based control for automotive systems, Electric and solar vehicles, Nonlinear and optimal automotive control
Abstract: Developing predictive control strategies is a time-consuming and expert-intense process, particularly for complex nonlinear systems. Classical predictive control approaches use first-principle models, which require significant effort for derivation, parameterization, and validation.

To mitigate these challenges, this work suggests the use of Neural Ordinary Differential Equations (NODEs) for fast and precise data-driven modeling. This approach provides a method for the automatic and easy design of models for predictive controllers, thereby meeting the industry's need for faster development, increased energy efficiency, and less effort. We demonstrate the efficiency and efficacy of this methodology in the application of fast and energy-optimal cabin cooling for a recent Battery Electric Vehicle (BEV) thermal system topology.

Keywords: AI and learning-based control for automotive systems, Engine and powertrain modeling and control, Adaptive and robust control of automotive systems
Abstract: To support decarbonization of transport, research is focusing on advanced high efficient combustion concepts running on low-carbon fuels. Optimizing the performance of these complex engines over a wide range of operating conditions results in exploded control calibration efforts for traditional calibration methods. In this paper, an automated engine calibration framework based on constrained Extremum Seeking (ES) control is proposed to effectively reduce calibration times. Contrary to previous research that optimizes an efficiency metric, the proposed ES algorithm directly shapes the entire in-cylinder pressure trace by decomposing it into Principal Components (PCs) and controlling the associated weights towards their optimal reference, defined by an Ideal Thermodynamic cycle. This method finds an optimal trade-off between controllability of the PC weights and optimality, while explicitly addressing combustion safety constraints using a novel gradient based projection method. The proposed ES controller was successfully implemented for fuel path calibration on a single cylinder engine running in dual-fuel (diesel-gasoline) mode, demonstrating optimality and convergence within 4 minutes and validating the handling of an arbitrary maximum in-cylinder pressure constraint. This work highlights the practical viability of automated calibration methods through a model-free in-cylinder pressure shaping approach.

Keywords: AI and learning-based control for automotive systems, Hybrid, electric and alternative drive vehicles, Modeling, supervision, control and diagnosis of automotive systems
Abstract: Future automotive powertrains are becoming increasingly complex, leading to exploding time and cost demands for calibration using conventional methods. This paper presents a Reinforcement Learning (RL)-based control strategy for a battery electric vehicle thermal system with safety constraints. A novel exploration approach combines an online Gaussian Process Regression model with a reciprocal Control Barrier Function to ensure safe, information-efficient learning. Validated on a vehicle test bench, the method autonomously optimizes heat pump steady-state operation under varying ambient conditions. The approach achieves heat pump efficiency within 2% of the optimum and reduces calibration time by 69% compared to conventional map-based methods.

Keywords: AI and learning-based control for automotive systems, Learning and adaptation in autonomous vehicles, Cooperative navigation
Abstract: Autonomous vehicles require accurate environmental sensing through onboard sensors, and their safety is compromised when those sensors fail to detect critical safety hazards. To address this issue, we present a control framework where a passenger acts as a cooperative safety monitor. Using natural language reports, such as ``there is an obstacle in front!'', a language understanding structure interprets the undetected safety issue and instantly generates a new virtual safety constraint in the vehicle's Model Predictive Control (MPC). This extension allows the vehicle to safely and in real-time avoid dangers not recognized by its own sensors. We also implement the framework in an autonomous vehicle with the CARLA simulator to demonstrate its effectiveness.

Keywords: Artificial intelligence in transportation, Information processing and decision support in transportation, Modeling and simulation of transportation systems
Abstract: Understanding and quantifying driving behavior is essential for enhancing fuel efficiency, reducing emissions, and improving fleet-level operational efficiency. This paper presents an interpretable deep learning framework for unsupervised driving regime discovery, the Temporal Convolutional Transformer with Spectral Clustering (TCTSC). The model integrates a Temporal Convolutional Network (TCN) for short-term vehicle dynamics, a Transformer encoder for long-range temporal dependencies, and spectral regularization with a differentiable K-Means module for compact and balanced clustering. The framework is trained and validated on large-scale Controller Area Network (CAN) data from Edmonton Transit Service diesel buses operating under diverse traffic and weather conditions. TCTSC achieves over 75% lower reconstruction errors than DenseAE and more than 85% lower than PCA, preserving both steady-state and transient behaviors. Outcome-grounded evaluation using fuel consumption shows that TCTSC explains 59.7% of the variance (R^2), achieves a Spearman correlation of 0.55, and attains the lowest MAE and RMSE among all models. The identified regimes, idling, deceleration, efficient cruising, high-load cruising, and aggressive acceleration, provide interpretable, physically meaningful insights into real-world bus operations and a robust foundation for data-driven eco-driving analysis and driver assessment.

Keywords: Modeling, supervision, control and diagnosis of automotive systems, Automotive system identification and modelling, Vehicle dynamic systems
Abstract: This study presents a robust method for detecting structural faults in vehicles using nonparametric frequency-domain identification. The approach uses a MIMO Local Rational Model to estimate transfer functions from accelerometer data. Physical experiments were conducted on a vehicle at various road surfaces, speeds, and induced fault conditions. Fault-specific features are extracted and classified using L1-regularized logistic regression. The proposed method achieved high detection accuracy in the experiments while being robust to changes in input excitation, demonstrating strong generalization performance.

Keywords: Applications of FDI/FTC, Control and management of energy systems, Fault detection and isolation methods
Abstract: The growing concern over climate change is driving the energy production research into renewable energy sources. Among them, wave energy stands out as a key player in complementing its sister renewables, i.e. solar and wind. Nonetheless, high commercialisation costs are hindering its widespread adoption. One of the key tools to reduce costs is constituted by energy-maximising optimal control strategies, which require accurate estimators of the excitation force for reaching energy generation optimality. In the state-of-the-art of excitation force estimation, there is a lack of algorithms capable of simultaneously handling modelling uncertainties, which are typical of experimental applications, without introducing additional assumptions about the nature of the waves, e.g. an explicit model of the excitation force itself. To fill this gap, this study proposes an excitation force estimator based on the implementation of a sliding mode observer, adapted from the field of fault diagnosis. The estimator, whose design is based on an identified model of the WaveStar prototype, is tested on experimental data collected under irregular sea state conditions to demonstrate its effectiveness in estimating the wave excitation force within a realistic framework.

Keywords: Fault detection and isolation methods, Applications of FDI/FTC
Abstract: Real-time knowledge of the excitation force (EF) that drives wave energy converters (WECs) has an important role in optimizing, and monitoring, the performance of these systems. corr{Since EF cannot be directly measured, estimation methods proposed in the literature are mainly based on (E1) observation of the WEC dynamics, or (E2) wave elevation measurements}. The accuracy of both types of estimators involves several factors, such as the complexity of the wave field and the physical structure of the WEC. This paper focuses on evaluating the potential of combining diverse EF estimates for fault diagnosis (FD) in WECs, since E1-type estimates are typically affected by changes in the system dynamics (faults), while E2-type estimates remain unaffected. The main contributions are the formulation of a fault detectability condition and the proposal of a scheme for detecting faults by processing different EF estimates. Numerical simulations of the WEC operation, in the presence of faults and parameter deviation in the WEC model, given a realistic sea wave profile, are performed to validate the proposed scheme.

Keywords: Applications of FDI/FTC, Fault detection and isolation methods, Wind power
Abstract: Effective fault detection and identification (FDI) is essential to reduce operation and maintenance costs in floating offshore wind turbines (FOWTs). To address this, the present paper considers a FOWT benchmark in the literature, proposing a Switching NARX scheme that alternates between a stable one-step predictor and a sensitive closed-loop simulator. The proposed FDI scheme incorporates a "backtracking" validation process, and a genetic algorithm is employed to automatically optimize the decision thresholds and logic parameters, maximizing the F1-score. Results demonstrate that this data-driven strategy provides robust detection with negligible computational cost, successfully identifying pitch sensor faults under realistic turbulent conditions.

Keywords: Applications of FDI/FTC, Fault detection and isolation methods, Wind power
Abstract: This paper provides a structured review of fault detection and isolation (FDI) methods across the offshore renewable energy (ORE) landscape, covering fixed-bottom offshore wind, floating offshore wind, wave energy converters, and tidal current turbines. For each technology, the study identifies the dominant fault types and target subsystems, analyzes the prevailing modeling approaches (model-based, data-driven, or hybrid), and assesses the current level of validation maturity, ranging from numerical simulation to field deployment. The analysis reveals a distinct correlation between the technology readiness level of each sector and its FDI focus: mature technologies prioritize grid-side reliability, while emerging sectors focus on structural integrity and primary conversion survivability. Finally, a cross-technology comparative analysis is presented, highlighting transferable methodologies, identifying the validation gap in data-driven strategies, and outlining emerging research priorities to enhance the resilience of future ORE systems.

Keywords: AI methods for FDI/FTC, Wind power
Abstract: By incorporating Fuzzy Neural Networks for on-the-fly damage estimation within a Linear Parameter-Varying framework, this paper proposes a health-aware Economic Model Predictive Control strategy for wind turbines. The proposed controller simultaneously optimises the economic rewards for turbine operation with the control of structural health, within the structural fatigue constraints of the blades. This is done to counterbalance the powering of blades, which is the primary cause of fatigue. A simulation with a 5 MW turbine model shows a 25% reduction in cumulative blade stress, along with a 15% increase in energy efficiency for this method, in comparison to standard PID control. This confirms the turbine maintains sustainable operation.

Keywords: Multi-agent systems, Distributed reinforcement learning, Control of networks
Abstract: Modern smart grids face critical challenges in coordinating distributed energy resources, demand response, and networked microgrids in a resilient, decentralized manner. This paper proposes a multi-agent reinforcement learning (MARL) framework for large-scale smart grid energy management, emphasizing decentralized optimization and control-theoretic rigor. The problem is formulated as a cooperative Markov game. A centralized-training, decentralized-execution (CTDE) MARL algorithm with actor–critic networks, attention mechanisms and prioritized experience replay is developed. Lyapunov-based safe RL ensures stability and safety throughout learning and deployment. Adversarial training and federated learning further enhance robustness and privacy. Simulation on a multi-microgrid test system shows an 18% cost reduction over stochastic programming, smoother power exchange, and no load shedding under attack.

Keywords: Distributed optimization for smart grids, Control and management of energy systems, Power systems stability
Abstract: We propose a method for online optimisation and control of DC microgrids with boost converters. In these systems, the control input is the switching duty cycle associated with each grid node. This leads to a bilinear term in the corresponding optimisation problem, which prevents the use of linear optimisation methods. To address this, we employ a recently proposed switched systems approach for online nonlinear optimisation of the microgrid, while guaranteeing convergence of the overall network. The effectiveness of the method is demonstrated through numerical simulation.

Keywords: Electrical distribution systems, Electrical transmission systems, Power systems stability
Abstract: The utilization of volatile energy sources will lead to challenges regarding voltage control in the future. Decentralized, inverter-based resources are capable of reactive power control, but are mainly installed in distribution grids. Integrating these resources in the voltage control of the transmission system presents a viable alternative to installing more compensating devices but requires coordination between distribution and transmission system operators. Such a control scheme must be able to account for operational limits in the distribution grid. This paper proposes a model predictive control (MPC)-based coordination scheme for the reactive power exchange at the point of common coupling between distribution and transmission systems that additionally accounts for reactive power reserves for dynamic voltage support (DVS) in distribution grids. The results show, that the coordination scheme effectively manages the reactive power exchange even while withholding reserves for DVS.

Keywords: Electrical distribution systems, Power systems stability, Real time simulators for energy systems
Abstract: Power system inertia reflects the grid’s capability to resist rapid frequency changes triggered by disturbances. The transition of electrical power systems generation from conventional fossil-fueled synchronous generators to converter-based renewable energy resources substantially reduces the inherent overall system inertia, which in turn impacts system stability and robustness. Therefore, in future power systems architectures, such as microgrids, it is vital to monitor the system inertia in real time. In this paper, we develop a Power-Hardware-in-the-Loop (PHiL) implementation of an online distributed inertia estimator for microgrids based on the consensus + innovations algorithm, which was originally developed for multi-area inertia estimation in synchronous generator-dominated power systems. Moreover, we show how the estimator performs under an actual 5G mobile communication architecture and compare the performance with that of a wired local area network, confirming its reliability and accuracy for real-time inertia estimation.

Keywords: Energy management systems, Electric vehicles integration in energy networks, Electric vehicles and charging stations
Abstract: The integration of photovoltaics (PV) with vehicle-to-grid (V2G)-capable electric vehicles (EVs) in residential microgrids offers a promising path toward sustainable energy man agement, yet faces challenges from the intermittent nature of PV generation and the stochastic, nonlinear dynamics of EVs. Existing results often rely on simplified EV models with constant charging/discharging efficiencies, limiting their practical efficacy, while high-fidelity modeling remains computationally prohibitive. To overcome these limitations, this paper proposes a robust data-driven control strategy based on zonotopic predictive control (ZPC) and mixed-integer programming (MIP) for coordinating multiple EVs in a PV-powered residential microgrid. The proposed strategy introduces multiple sets of binary variables to prevent simultaneous charg ing/discharging of multiple EVs and employs a matrix-zonotope recursion for over-approximated reachable-set computation, thereby ensuring robust constraint satisfaction under unknown but bounded disturbances. Numerical simulations demonstrate that the proposed strategy effectively enhances PV self-consumption and reduces grid electricity costs, lowering the average PV curtailment rate from 8.13% to 3.26% while maintaining operational reliability.

Keywords: Power plant control, Power systems stability, Distributed optimization for smart grids
Abstract: With the growing penetration of converter-interfaced renewable energy resources, virtual power plants (VPPs) have become an effective framework for coordinating heterogeneous distributed energy resources (DERs). To provide ancillary services in a power system, a VPP must deliver a time-varying power response, such as active power adjustments driven by system frequency deviations. The optimal allocation of the individual DER contributions to the overall VPP response can be cast as a time-varying constrained optimization problem. To address this problem, we employ a finite-time primal-dual gradient descent (FT-PDGD) algorithm, whose convergence to the optimal time-varying trajectory is formally established through a Lyapunov-based analysis. Experimental validation on a Power Hardware-in-the-Loop testbed demonstrates the real-time performance of the proposed FT-PDGD. Furthermore, we compare its performance with two existing FT-PDGD variants from the literature and the standard PDGD algorithm.

Keywords: Power systems stability, Wind power, Electrical transmission systems
Abstract: The decline of inertia in modern power systems jeopardizes frequency stability. Offshore wind farms (OWFs) connected via high-voltage direct current (HVDC) links can mitigate this by providing virtual inertia, but inertia provision conflicts with DC-voltage stability of the link. This paper proposes a modular, communication-free control scheme that trades off both aspects. Onshore, a grid-forming converter with virtual synchronous machine behaviour is combined with PID DC-voltage control and a limited grid-forming loop. Offshore, a DC-voltage-to-frequency law interacts with a grid-following OWF with frequency-sensitive mode control operated without curtailment. Linearised Laplace-domain analysis clarifies impacts on inertia and damping. Time-domain simulations confirm improved frequency support, bounded DC-voltage deviations, and fast post-fault power realignment.

Keywords: Cyber physical systems, Safety and security in networked control, Control software architecture
Abstract: Cyber-physical systems (CPS) are increasingly utilized in various applications including smart factories, autonomous driving, and healthcare, where secure processing of sensitive data is critical. Fully homomorphic encryption (FHE), which has emerged as a core technology for enhancing CPS security, enables computation over encrypted data, preventing exposure of sensor outputs, state estimates, and control inputs to potential adversaries. However, the practical deployment of FHE in control systems faces significant challenges due to computational overhead, especially from encrypted multiplication, resulting in high latency. In this work, we propose an efficient scalar-ciphertext multiplication method that employs primitive gates together with a variant of the non-adjacent form (NAF), thereby significantly reducing the computational cost. Simulation results show that the proposed method reduces the scalar-ciphertext multiplication latency by up to 63.9% compared to conventional FHE-based schemes.

Keywords: Safety and security in networked control
Abstract: While computation-enabled cryptosystems applied to control systems have improved security and privacy, a major issue is that the number of recursive operations on encrypted data is limited to a finite number of times in most cases, especially where fast computation is required. To allow for nonlinear dynamic control under this constraint, a method for representing a state-space system model as an auto-regressive model with exogenous inputs (ARX model) is proposed. With the input as well as the output of the plant encrypted and transmitted to the controller, the reformulated ARX form can compute each output using only a finite number of operations, from its several previous inputs and outputs. Existence of a stable observer for the controller is a key condition for the proposed representation. The representation replaces the controller with an observer form and applies a method similar to finite-impulse-response approximation. It is verified that the approximation error and its effect can be made arbitrarily small by an appropriate choice of a parameter, under stability of the observer and the closed-loop system. Simulation results demonstrate the effectiveness of the proposed method.

Keywords: Safety and security in networked control, Cyber physical systems
Abstract: This paper proposes an encrypted state observer that is capable of detecting sensor attacks without decryption. We first design a state observer that operates over a finite field of integers with the modular arithmetic. The observer generates a residue signal that, under sparse attack and sensing redundancy conditions, indicates the presence of attacks. Then, we develop a homomorphic encryption scheme that enables the observer to operate over encrypted data while automatically disclosing the residue signal. Unlike our previous work restricted to single-input single-output systems, the proposed scheme is applicable to general multi-input multi-output systems. Given that the disclosed residue signal remains below a prescribed threshold, the full state can be recovered as an encrypted message.

Keywords: Safety and security in networked control, Cyber physical systems, IT/OT-security in automation systems
Abstract: A secure two-party computation protocol for running dynamic controllers over secret sharing has recently been proposed. Unlike encrypted control schemes based on homomorphic encryption, this protocol enables operating dynamic controllers for an infinite time horizon without controller-state decryption, controller-state reset, or input re-encryption. However, the two-party setting introduces additional online communication between the computing parties, which may hinder real-time feasibility. In this study, we demonstrate the feasibility of the protocol through implementation on a commercial cloud platform with an inverted pendulum testbed. Experimental results show that the proposed protocol successfully stabilized the pendulum despite the online communication overhead.

Keywords: Safety and security in networked control, Cyber physical systems, Virtualized and cloud-based control architectures
Abstract: The security of networked control systems (NCS) is receiving increasing attention from both cyber-security and system-theoretic perspectives. The former focuses on classical IT security goals such as confidentiality, integrity, and availability of process data, while the latter investigates tailored attacks (and detection schemes), including covert and zero-dynamics attacks. Confidentiality in control systems can, for instance, be achieved by securely outsourcing the evaluation of the controller to third-party platforms, such as cloud services. The underlying technology enabling such secure computation often is homomorphic encryption (HE).

Recent works in encrypted control have proposed modifications to underlying HE schemes to achieve not only confidentiality but also resilience to certain types of integrity attacks. While extensions in this direction are desirable in principle, we show that the integrity problem in encrypted control cannot be solved by public-key HE schemes alone due to their inherent malleability. In other words, the same homomorphisms that enable encrypted control % in the first place can be leveraged not only constructively but also destructively. More precisely, we demonstrate that NCS are vulnerable to covert attacks, even when encrypted control is employed. Remarkably, this remains possible without knowledge of an unencrypted model.

Yet, resilience to such attacks can still be achieved through complementary techniques. We present an approach based on verifiable computation that integrates with modern homomorphic cryptosystems and is asymptotically secure while incurring no communication overhead.

Keywords: Sensing and perception in agriculture, Agricultural robotics, Positioning and navigation in agriculture and forestry
Abstract: Autonomous roadside mulching requires perception and planning systems capable of reasoning about the 3D structure of the environment, which consists of vegetation and roadside objects such as poles, traffic signs, and guardrails. This paper addresses the problem by investigating and applying Normal Distribution Transform (NDT)–based traversability estimation for perception. In addition, two one-class support vector machines (SVMs) are trained to separately estimate the road surface and traversable regions in the roadside area. For planning, a path planner based on dynamic programming is proposed that uses traversability estimates and the road edge to select a collision-safe Cartesian path for a roadside mulcher. The effectiveness of the proposed system is demonstrated in real Finnish roadside conditions. Preliminary results show that the system can reliably detect objects such as utility poles in roadside environments and generate Cartesian paths that avoid collisions. In future work, the resulting Cartesian paths will be integrated into an automatic nonlinear model predictive control (NMPC) framework for full autonomous operation.

Keywords: Positioning and navigation in agriculture and forestry, Agricultural robotics, Control in precision agriculture
Abstract: Guiding a tractor along a predefined reference path is a key component of precision agriculture. This study develops a path tracking controller based on Nonlinear Model Predictive Control, which incorporates multiple segments of a piecewise-linear reference path directly into the objective function. In addition, methods for selecting viable reference segments from the full path are presented. The control system is evaluated during a field test with a tractor controlled via the Tractor Implement Management steering interface. The NMPC solver converged on average after 3.45 ms and tracked the curved reference path with a mean absolute cross-track error of 6.1 cm.

Keywords: Modeling and estimation in agriculture, Agricultural robotics, Control in precision agriculture
Abstract: This paper presents a mathematical model and an intuitive Cartesian joystick controller for a tractor-mounted hydraulic boom mulcher used in roadside vegetation management. The boom is modeled as a four-degree-of-freedom manipulator with revolute joints actuated by hydraulic cylinders. The kinematic model combines Denavit--Hartenberg parameters, geometric constraints, and analytical and numerical mappings between cylinder lengths and joint angles. Cylinder dynamics are represented by first-order models with range limits. A Jacobian-based resolved-rate motion controller maps Cartesian joystick commands to joint velocities, which are converted to cylinder velocities for simulation. Cylinder limits are respected using constraint-aware task scaling to ensure smooth motion near limits. The resulting virtual prototype, implemented in MATLAB and RViz with ROS 2, enables intuitive end-effector control and provides a foundation for future nonlinear model predictive control aimed at reducing operator workload and improving safety in roadside vegetation management.

Keywords: Agricultural robotics, Positioning and navigation in agriculture and forestry
Abstract: This paper presents a new method for headland coverage path planning for arable fields. Several earlier approaches suggest covering the headland with nested polygons and smooth turns, however, covering the field corners entirely requires manoeuvres with reversing. In the new method, the polygon corners are modified to allow a reversing turn. A comparison to two other methods considering gap, overlap, and crossing the field boundary shows an improvement in the coverage result especially in field corners of around 90 degrees, and 240 degrees and above. Applicability of the new method is shown with several examples of real polygonal field maps.