Keywords: Robotic learning and adaptation
Abstract: The acrobatic feats of recent humanoids notwithstanding, today's robots still struggle with everyday tasks like opening jars, inserting plugs, using scissors, and other manipulation tasks that humans perform effortlessly. While hardware limitations are partly to blame — today’s robot hands still lack the strength, precision, flexibility, and sensing capabilities of human hands — it is becoming increasingly clear that traditional model-based control methods are also inadequate. Manipulation involves the coordinated control of motion, force, and compliance under uncertainty, disturbances, unmodeled dynamics, and physical constraints imposed by the task and environment. During manipulation a robot may shift between open- and closed-chain systems, and between underactuated and overactuated regimes.

Data-driven methods offer a promising alternative to manually engineering such complex control laws. However, current approaches based on Vision-Language-Action (VLA) models have major shortcomings: they are often device-dependent, require enormous training data, scale poorly, and struggle to generalize across tasks. More fundamentally, they fail to fully leverage the extensive body of knowledge accumulated in control systems design and the mechanics of manipulation.

This talk proposes a new control architecture for robot manipulation that merges key concepts from control theory, mechanics, geometry, and human motor control with data-driven methods. We show that Brockett's motion description language (MDL) paradigm provides a device-independent, hierarchical, and modular framework for manipulation expressed in the language of control systems. At the highest level, foundation models decompose complex tasks into subtasks, which are then encoded as sequences of robot action primitives using both state-space control and data-driven learning methods. At lower levels, we show how coordinate-invariant geometric methods can be used to construct minimum distortion latent space manifolds, equivariant learning models, and minimum attention feedforward-feedback control laws.

Keywords: Linear system identification
Abstract: The objective of this short tutorial is to explain why any jointly controllable, jointly observable multi-channel linear system with a strongly connected neighbor {communication} graph can be exponentially stabilized with arbitrarily fast convergence using a time-invariant distributed linear control. This fact can be established in several different ways. One way involves using a distributed observer-based control architecture analogous to the familiar centralized observer-based architecture used to control a controllable, observable, linear system. Distributed control can be thought of as a generalization of decentralized control in which communication between neighboring agents is allowed. An important consequence of this generalization is that the well-known fixed spectrum {set of fixed modes} of a linear system which arises with decentralized control is no longer an obstacle to the system's regulation with distributed rather than decentralized control. Perhaps the most important idea in automatic control is the concept of integral control. We will explain how to use integral control in a distributed setting to realize a feedback control which enables each and every agent with access to the system to independently adjust its controlled output to any desired set-point value. Often overlooked in the study of distributed control are the effects of communication network transmission delays. It will be explained why in the face of such delays, exponential stabilization at a prescribed convergence rate can still be achieved with distributed control, at least for discrete-time, multi-channel linear systems.

Keywords: Linear system identification
Abstract: This short tutorial discusses resilience challenges in distributed optimization when some agents in the network behave adversarially and provide unreliable information. We focus on Byzantine settings, where compromised agents can arbitrarily manipulate messages exchanged over the communication network. To address this challenge, we introduce resilience design principles based on graph redundancy and objective redundancy, which enable reliable coordination despite the presence of adversarial agents. Using distributed subgradient methods as an illustrative example, we show how these principles ensure that all non-adversarial agents asymptotically agree on an optimal solution under suitable conditions, and we briefly discuss insights into convergence rates. The talk further demonstrates that the same resilience ideas extend beyond optimization to other distributed computational tasks, including distributed linear equation solving, leading to fully resilient algorithms that tolerate Byzantine behavior. The goal is to convey general design insights that apply across a broad class of distributed control and optimization problems.

Keywords: Linear system identification

Keywords: Linear system identification

Keywords: Application of nonlinear analysis and design, Control barrier functions and state space constraints, Nonlinear model reduction
Abstract: In this paper, we investigate the problem of differential flatness-based two-layer control strategy for fixed-wing autonomous aerial vehicles (AAVs). Firstly, the dynamics of fixed-wing AAVs is transformed through differential flatness, where all states and inputs are denoted as the functions of flat outputs and their derivatives. Based on this transformation, the multiple constraints existed in practical flights can be unified to the constraints on flat outputs. This ensures that the inherent connections among constraints are fully regarded, and the propagation of constraints occurred in dynamics of fixed-wing AAVs is resolved. Then, we design a two-layer control strategy, consisting of control commands (accelerations) and actual controllers (thrust and control surfaces). It balances the stability and practical feasibility for fixed-wing AAVs. Finally, a simulation is conducted to verify the effectiveness of the proposed method in an obstacle environment.

Keywords: Application of nonlinear analysis and design, Optimization-based estimation and control
Abstract: This paper develops a smooth safety-filtering framework for nonlinear control-affine systems under limited perception. Classical Control Barrier Function (CBF) filters assume global availability of the safety function---its value and gradient must be known everywhere---an assumption incompatible with sensing-limited settings, and the resulting filters often exhibit nonsmooth switching when constraints activate. We propose two complementary perception-aware safety filters applicable to general control-invariant safety sets. The first introduces a smooth perception gate that modulates barrier constraints based on sensing range, yielding a closed-form Lipschitz-safe controller with forward-invariance guarantees. The second replaces the hard CBF constraint with a differentiable penalty term, leading to a smooth unconstrained optimization-based safety filter consistent with CBF principles. For both designs, we establish existence, uniqueness, and forward invariance of the closed-loop trajectories. Numerical results demonstrate that the proposed smooth filters enable the synthesis of higher-order tracking controllers for systems such as drones and second-order ground robots, offering substantially smoother and more robust safety-critical behaviors than classical CBF-based filters.

Keywords: Control barrier functions and state space constraints
Abstract: This paper proposes a set-relaxed disturbance-resistant high-order control barrier function (SRDR-HOCBF) frameworks to address limitations in existing robust CBF methods under parameter uncertainties and external disturbances. The framework employs a recursive virtual constraint relaxation mechanism to systematically enlarge the forward invariant set, and theoretical proofs establish the forward invariance under bounded uncertainties and disturbances. Comparative simulations on a horizontal pendulum and a mobile navigation system validate its superiority in safety maintenance over traditional HOCBF. And it is particularly effective in reducing initial state requirements, outperforming other methods under broader initial conditions when integrated with control strategies.

Keywords: Control barrier functions and state space constraints, Adaptive control design
Abstract: Control barrier functions (CBFs) have proven effective in guaranteeing the safety of control systems; however, accurate system model is usually required for CBF-based controller design, which is generally difficult to obtain in practice. While uncertainty estimation and compensation can enhance robustness of CBFs, existing methods typically need the bounds of uncertain term to reject residual estimation error. This paper considers a more complex scenario where the system is subject to completely unknown parametric uncertainties, including both measurement errors and parametric deviations. Such compound uncertainty poses significant challenge for existing CBF approaches, which require the bounds of both measurement error and the parameter deviation to guarantee safety. To overcome this limitation, we propose a novel class of CBFs, called the reciprocal-compensated uncertainty-aware CBF, to enforce robust safety against uncertainties without requiring any prior knowledge of these uncertainties. A simulation example demonstrates the effectiveness of our approach.

Keywords: Control barrier functions and state space constraints, Application of nonlinear analysis and design
Abstract: This paper proposes an integral control barrier function (I-CBF)-based safety augmentation method for aircraft trajectory management with static obstacle avoidance. We consider a point-mass model of a cruising aircraft in which thrust, bank angle, and flight-path angle are inputs, while a waypoint-based guidance law and low-level proportional controllers define input dynamics. To handle the position-based safety constraint within the I-CBF framework, we define a barrier function as the minimum safety margin to the obstacle over a short-horizon predicted trajectory. The required gradients with respect to the current state and input are computed by integrating sensitivity equations along the prediction. This yields a linear constraint on the auxiliary input and a small quadratic programming, which can be incorporated into the I-CBF framework. Simulation using a Boeing 787-8 model shows that the proposed safety augmentation keeps the aircraft away from the static obstacle with only small deviations from the nominal waypoint-tracking path.

Keywords: Control barrier functions and state space constraints, Application of nonlinear analysis and design, Optimal control theory
Abstract: For nonlinear affine systems affected by unknown disturbances and input constraints, this study develops a safety critical control method based on higher-order control barrier functions(HoCBF). Firstly, to suppress the persistent impact of unknown disturbances on the safety constraint performance, a disturbance observer-based tunable input-to-state-safe HoCBF is designed, further reducing the conservatism of the safety constraints. Secondly, a time-varying function is incorporated into the construction of the HoCBF to address input constraints in safety critical control. By designing an auxiliary dynamic system to dynamically adjust the safety set, the conflict between input saturation and safety constraints is mitigated, effectively preventing infeasible solutions in quadratic programming problems under multiple constraints. Finally, the effectiveness and superiority of the suggested framework are validated via experiments on unmanned ground vehicles cooperative obstacle avoidance.

Keywords: Control barrier functions and state space constraints, Controller constraints and structure, Application of nonlinear analysis and design
Abstract: Quadratic Programs (QP) subject to Control Barrier Function (CBF)-based constraints are widely employed to design safety-critical controllers. However, ensuring the feasibility of the QP under input constraints remains a significant challenge. In this work, we propose a feasibility-margin-based CBF as a proactive safety filter to guarantee the dynamic feasibility of CBF-QP with input constraints. We first characterize the feasibility margin using support functions defined by the geometry of input constraints. We then propose a novel safe control method that employs the feasibility margin as a Control Barrier Function (FMA-CBF) for safety-critical control systems subject to polytopic input constraints. Furthermore, we formulate a unified QP that enforces both the original safety constraints and the feasibility margin constraints to guarantee feasibility. The efficacy of the proposed method is validated through double-integrator systems and unicycle robots with obstacle avoidance tasks.

Keywords: Control barrier functions and state space constraints, Controller constraints and structure, Application of nonlinear analysis and design
Abstract: This paper addresses the safe navigation problem for a unicycle-type mobile robot operating in obstacle-cluttered environments. Existing safe control approaches typically employ control barrier functions (CBFs) to formulate a quadratic programming (QP) problem that minimally modifies a given nominal control input to ensure safety. However, within this CBF-QP framework, the direct application of high-order or hybrid-order CBFs to unicycle-type robots remains limited in practicality. To overcome this limitation, we first analyze the relative position dynamics between the robot and obstacles and develop a novel safe control method using a first-order CBF. This formulation enables effective obstacle avoidance based directly on point cloud data from an onboard LiDAR. Furthermore, to alleviate the computational burden associated with processing dense point clouds, we propose an efficient point cloud filtering strategy that significantly reduces the number of CBF constraints in the QP without compromising safety. Finally, the efficacy of the proposed method is validated on the NVIDIA IsaacSim platform.

Keywords: Control barrier functions and state space constraints, Controller constraints and structure, Optimization-based estimation and control
Abstract: This paper considers trajectory tracking control and collision avoidance for linear multi-agent systems (MAS) with bounded input constraints based on the control barrier function (CBF) design. We identify the conflict between leader tracking performance and follower control freedom in input-constrained multi-agent systems. And then propose a uniformly time-varying CBF to cope with the state constraints. Moreover, the trade-off between the control freedom of the leader and follower agents is examined. Conservativeness about the satisfaction of the constraints is quantified as a condition on the selection of the function used for the controller design. Some simulations are provided to illustrate the effects of the virtual leader actuation on the swarm of the follower agents and to demonstrate the efficacy of our design.

Keywords: Control barrier functions and state space constraints, Convex optimization
Abstract: This paper presents a novel feasible-set reshaping technique to optimization-based control with ensured constraint qualification. In our problem setting, the feasible set of admissible control inputs depends on the real-time state of the plant, and the linear independence constraint qualification (LICQ) may not be satisfied in some regions of interest. By feasible-set reshaping, we project the constraints of the original feasible set onto an appropriately chosen constant matrix with its rows forming a positive span of the space of the optimization variable. It is proved that the reshaped feasible set is nonempty and satisfies LICQ, as long as the original feasible set is nonempty. The effectiveness of the proposed method is verified by constructing Lipschitz continuous quadratic-program-based controllers based on the reshaped feasible sets.

Keywords: Control barrier functions and state space constraints, Decentralized control
Abstract: Control Barrier Function based Quadratic Programs (CBF-QPs) are widely used for collision avoidance in multi-robot systems, but their real-time implementation is limited by the computational cost of online optimization. Recently, Control Barrier Proximal Dynamics (CBPD) reformulates CBF-QPs as continuous-time dynamics and offers significant computational speedups. However, existing results are restricted to affine constraints and cannot handle the nonlinear quadratic constraints arising in collision avoidance. This paper proposes a Collision Avoidance-CBPD (CA-CBPD) framework. We establish strong contraction under a time-varying metric and prove that its tracking error with respect to the QP solution remains uniformly bounded. The maximum safety violation is explicitly quantified, enabling a robust compensation strategy with guaranteed safety. Numerical results show that CA-CBPD achieves over 200× speedup compared with CBF-QP while maintaining reliable collision avoidance.

Keywords: Control barrier functions and state space constraints, Disturbance rejection and input-to-state stability, Stability of nonlinear systems
Abstract: This paper develops a robust safety-critical control method for nonlinear strictfeedback systems with mismatched disturbances. Using a state transformation and a linear time-varying disturbance observer, the system is converted into a form that enables safe control design. The approach ensures forward invariance of the safety set and also applies to disturbancefree systems. Safety is proven for all cases, and a numerical example illustrates the results.

Keywords: Control barrier functions and state space constraints, Model predictive control, Optimal control theory
Abstract: Optimal control strategies are often combined with safety certificates to ensure both performance and safety in safety-critical systems. A prominent example is combining Model Predictive Control (MPC) with Control Barrier Functions (CBF). Yet, efficient tuning of MPC parameters and choosing an appropriate class Kappa function in the CBF is challenging and problem dependent. This paper introduces a safe model-based Reinforcement Learning (RL) framework where a parametric MPC controller incorporates a CBF constraint with a parameterized class Kappa function and serves as a function approximator to learn improved safe control policies from data. Three variations of the framework are introduced, distinguished by the way the optimization problem is formulated and the class Kappa function is parameterized, including neural architectures. Numerical experiments on a discrete double-integrator with static and dynamic obstacles demonstrate that the proposed methods improve performance while ensuring safety.

Keywords: Control barrier functions and state space constraints, Model validation, Learning methods for optimal control
Abstract: Formal verification of neural control barrier functions (NCBFs) remains challenging, especially for neural networks with nonlinear activations like tanh. Existing CROWN- based methods rely on conservative linear relaxations for Jacobian bounds, limiting scal- ability. We propose LightCROWN, which computes tighter Jacobian bounds by exploit- ing the analytical properties of activation functions. Experiments on nonlinear control sys- tems including the inverted pendulum, Dubins car, and planar quadrotor demonstrate that LightCROWN improves verification success rates up to 100%, while enhancing speed and scalability. Our approach provides a generalizable improvement for CROWN-based frameworks,enabling more efficient verification of complex NCBFs. The code can be found at github.com/ Autonomous-Systems-and-Control-Lab/verify-neural-CBF.

Keywords: Control barrier functions and state space constraints, Nonlinear observers and filters, Output regulation and tracking
Abstract: This paper presents an integrated safe-tracking control scheme for high-relative-degree nonlinear systems with uncertain dynamics and partial measurements. A Gaussian process (GPs) model and a high-gain observer jointly estimate the full state and learn unknown dynamics, with convergence of both estimation errors under suitable gain conditions. GP-based learning alleviates the need for large observer gains, mitigating peaking. Exponential control Lyapunov and barrier functions embedded in a one-step optimization-based controller with probabilistic guarantees enforce safety and tracking while prioritizing safety. Simulations show safe outputs, improved tracking, smoother inputs, and reduced observer gains versus GP-adapted with higher observer gains and observer-only approaches.

Keywords: Control barrier functions and state space constraints, Observer design
Abstract: Safety assurance for autonomous systems is challenged by unmatched disturbances, especially those with non-differentiable components like sensor noise. Existing methods are either incapable of dealing with such noise or are overly conservative. This paper proposes a novel disturbance observer-based disturbance rejection control barrier function framework for high-relative-degree safety constraints under composite disturbances. We integrate a disturbance observer with a robust disturbance rejection law to achieve less conservative performance while guaranteeing safety. Theoretical analysis and simulation study demonstrate that the proposed method guarantees safety under the unmatched composite disturbances, while outperforming a state-of-the-art robust approach.

Keywords: Control barrier functions and state space constraints, Optimization-based estimation and control, Numerical methods for optimal control
Abstract: Control barrier functions (CBFs) provide safety filters for constrained systems, but synthesizing a useful CBF can be difficult when the safe set is nonconvex or poorly represented by a prescribed function class. This paper develops a sampled-data, table-based CBF synthesis framework that uses finite-state prediction rather than a fixed analytic parametrization. The method evaluates each grid state through an instantaneous safety capacity, which measures the fraction of admissible inputs that are one-step safe, and an expected safety horizon, which accumulates this capacity along predicted sampled trajectories. The resulting update distinguishes states that are immediately feasible but have poor future recoverability from those with longer-term safety margins. Dubins car obstacle-avoidance simulations illustrate the construction of non-polynomial safe sets in cluttered environments and compare the result with a polynomial SOS-CBF baseline.

Keywords: Control barrier functions and state space constraints, Output feedback nonlinear control
Abstract: Control Barrier Functions (CBFs) provide a powerful framework for ensuring safety in dynamical systems. However, their application typically relies on full state information, which is often violated in real-world due to the availability of partial state information. In this work, we propose a neural network-based framework for the co design of a safety controller, observer, and CBF for partially observed continuous-time systems with input constraints. By formulating barrier conditions over an augmented state space, our approach ensures safety without requiring bounded estimation errors or handcrafted barrier functions. All components are jointly trained by formulating appropriate loss functions, and we introduce a validity condition to provide formal safety guarantees beyond the training data. Finally, we demonstrate the effectiveness of the proposed approach through several case studies.

Keywords: Control barrier functions and state space constraints, Uncertain systems, Analytic design
Abstract: This paper studies satisfying temporal logic specifications on stochastic dynamical systems, where the predicates evolve randomly over time. Such randomness may arise from uncertain environment models or external stochastic processes causing the sets associated with predicate satisfaction to vary in a non-deterministic manner. As a result, verifying whether a stochastic dynamical system satisfies a temporal specification depends also on the uncertainty in the predicates. We develop a certificate-based framework to bound the probability of satisfying temporal logic specifications with randomly evolving predicates. We first show that temporal logic specifications with stochastic predicates can be transformed to specifications with deterministic predicates on an augmented space which is extended to include the stochastic space of predicate’s uncertainty. We then utilize barrier certificates on an augmented space to provide tractable optimization-based conditions and to avoid the computational burden of dynamic programming. Focusing on linear dynamics and safety-type specifications, we derive analytical conditions under which barrier certificates guarantee bounds on the probability of violating the stochastic safety predicates. The approach is demonstrated on numerical case studies.

Keywords: Control barrier functions and state space constraints, Uncertain systems, Lyapunov methods
Abstract: We study the prescribed-time reach-avoid (PT-RA) control problem for nonlinear systems with unknown dynamics operating in environments with moving obstacles. Unlike robust or learning-based Control Barrier Function (CBF) methods, the proposed framework re- quires neither online model learning nor uncertainty bound estimation. A CBF-based Quadratic Program (CBF-QP) is solved on a simple virtual system to generate a safe reference satisfying PT-RA conditions with respect to time-varying, tightened obstacle and goal sets. The true system is confined to a Virtual Confinement Zone (VCZ) around this reference using an approximation-free feedback law. This construction guarantees real-time safety and prescribed- time target reachability under unknown dynamics and dynamic constraints without explicit model identification or offline precomputation. Simulation results illustrate reliable dynamic obstacle avoidance and timely convergence to the target set.

Keywords: Lyapunov methods
Abstract: Providing finite-time probabilistic safety and reach-avoid guarantees is crucial for safety-critical stochastic systems. Existing barrier certificate methods often rely on a restrictive boundedness assumption for auxiliary functions, limiting their applicability. This paper presents refined barrier-like conditions that remove this assumption. Specifically, we establish conditions for deriving upper bounds on finite-time safety probabilities in discrete-time systems and lower bounds on finite-time reach-avoid probabilities in continuous-time systems. This key relaxation significantly expands the class of verifiable systems, especially those with unbounded state spaces, and facilitates the application of advanced optimization techniques, such as semi-definite programming with polynomial functions. The efficacy of our approach is validated through numerical examples.

Keywords: Nonlinear control of switched & hybrid systems
Abstract: This paper investigates forward-invariant control of switched systems, allowing different subsystems to possess different safe sets. By analyzing the influence of subsystem dynamics and switching signals on the forward invariance of the safe set, a relaxed safety condition for individual subsystems is proposed. This condition requires the sub-tangential condition to hold only on a subregion of the safe set, rather than on the entire set. Consequently, individual subsystems may be unsafe, while overall system safety is achieved through switching control. Based on these relaxed safety conditions, an extended Nagumo’s theorem is established within a switched-systems framework. A clear and intuitive proof is provided for the practical safe sets commonly used in engineering, without relying on nontrivial tools from topology or functional analysis. In a special case, a necessary and sufficient condition is provided for the forward invariance of the safe set under arbitrary switchings. Finally, a compass-like biped walking robot example is used to demonstrate the effectiveness of the proposed method.

Keywords: Passivity-based control, Controller constraints and structure
Abstract: Krasovskii passivity is a passivity property defined by selecting the time derivative of the input as the input port variable. Because of this structure, Krasovskii passivity naturally yields integral controllers which are Krasovskii passive. In this paper, we show that such integral control schemes can easily be adapted to handle input bound constraints. Our approach consists of passing the inputs through activation-like functions and modifying the controllers so as to preserve their Krasovskii passivity. We apply the proposed tailoring method to stabilization, output consensus, and input consensus problems. For consensus controllers, we additionally demonstrate how slope constraints on the inputs can be enforced.

Keywords: Passivity-based control, Learning methods for optimal control
Abstract: Interconnection and Damping Assignment Passivity-Based Control (IDA-PBC) is a nonlinear control technique that assigns a port-Hamiltonian (pH) structure to a controlled system using a state-feedback law. While IDA-PBC has been extensively studied and applied to many systems, its practical implementation often remains confined to academic examples and, almost exclusively, to stabilization tasks. The main limitation of IDA-PBC stems from the complexity of analytically solving a set of partial differential equations (PDEs), referred to as the matching conditions, which enforce the pH structure of the closed-loop system. However, this is extremely challenging, especially for complex physical systems and tasks. In this work, we propose a novel numerical approach for designing IDA-PBC controllers without solving the matching PDEs exactly. We cast the IDA-PBC problem as the learning of a neural ordinary differential equation. In particular, we rely on sparse dictionary learning to parametrize the desired closed-loop system as a sparse linear combination of nonlinear state-dependent functions. Optimization of the controller parameters is achieved by solving a multi-objective optimization problem whose cost function is composed of a generic task-dependent cost and a matching condition-dependent cost. Our numerical results show that the proposed method enables (i) IDA-PBC to be applicable to complex tasks beyond stabilization, such as the discovery of periodic oscillatory behaviors, (ii) the derivation of closed-form expressions of the controlled system, including residual terms in case of approximate matching, and (iii) stability analysis of the learned controller.

Keywords: Robotic grasping and manipulation, Mechatronic system modeling, design, optimization, Mechatronic system estimation, identification, control
Abstract: In this work, we revisit the use of the filtered inverse algorithm to address multi-objective control of robot manipulators. The method employs the concept of dynamic inversion of the Jacobian matrix to handle kinematic singularities and augmented task-space problems, which may be ill-posed and involve conflicting goals. Herein, we evaluate different approaches for incorporating both trajectory tracking and the additional control objective of velocity manipulability maximization, as it correlates with the energetic efficiency of robotic operations. Finally, numerical simulations of a redundant planar arm demonstrate the behavior and performance of the proposed solution.

Keywords: Robotic grasping and manipulation, Task and motion planning
Abstract: This paper investigates the use of physics simulation for the motion planning of deformable linear objects (DLOs) like cables and ropes. Existing work has largely focused on the modeling of equilibrium configurations in such a way that standard sampling-based planners can be applied. However, these methods are difficult to extend to scenarios that require or allow contact between the DLO and the environment. Therefore, it seems attractive to directly use physics simulations like MuJoCo for the planning process instead of relying on equilibrium models. Concepts for lattice-based and tree-based planning are presented and compared to a state-of-the-art model for an intentionally simplified task to highlight advantages and challenges.

Keywords: Robotic learning and adaptation, Adaptive and adaptable automation, Medical and rehabilitation robotics
Abstract: In fields such as rehabilitation and biomechanics, many robotic systems have been developed to interact directly with humans. However, the speed of human movement is not constant for a given task, as the time required to complete an action varies with individual decisions. To address this variability, we develop a spatially based event controller that adapts to unknown parameters while allowing for movements in multiple directions, addressing limitations of existing spatial controllers. We derive the conditions on the controller parameters and event design that ensure system stability, and then present simulation examples demonstrating the controller’s ability to track spatial paths without constraining the terminal time.

Keywords: Robotic learning and adaptation, Autonomous navigation, Robot perception and sensing
Abstract: Simultaneous localization and mapping (SLAM) and navigation are core capabilities for agents, yet traditional methods rely on high-precision sensors and perform poorly in rapidly changing large-scale environments. Inspired by mammals' spatial cognitive and navigation mechanisms in neuroscience, this paper proposes a novel brain-inspired computational network for learning cognitive map representations and navigation in unknown environments. The network model diverse spatial cells to integrate perception and motion information for environmental representation, establishes a dynamically growing place cell-based cognitive map, and updates synaptic strength between place cells via agent-environment interaction to restructure the map. Additionally, a place cell sequence planning algorithm is designed for navigation using the cognitive map as input. Simulation and physical-robot experiments show that the proposed method can dynamically construct and update cognitive maps during environmental interaction and can improve navigation efficiency in the tested dynamic scenarios. These results suggest a feasible brain-inspired alternative for map learning and navigation in unknown environments.

Keywords: Robotic learning and adaptation, Teleoperation, AI-powered robotics
Abstract: Stochastic communication delays in teleoperation introduce signal discontinuities that undermine control stability and degrade control performance. Consequently, the conventional reinforcement learning (RL) methods struggle with the delayed observations due to the delay-induced observations, leading to high-frequency chattering. To address this, we propose a hybrid control framework, delay-resilient RL, integrating a state estimator utilizing Long Short-Term Memory (LSTM) with a residual RL policy, which is resilient to stochastic delays. The LSTM reconstructs smooth, continuous state estimates from delayed observations, enabling the RL agent to learn a residual torque compensation policy that balances tracking accuracy with velocity smoothness. Experimental validation on Franka Panda robots demonstrates that our approach significantly outperforms the state-of-the-art baselines, ensuring robust and stable teleoperation even under high-variance stochastic delays.

Keywords: Security for stochastic systems
Abstract: Stealthy attacks and defenses in mobile systems have been studied as zero-sum games, where an attacker covertly drives the system to an unsafe region and a defender detects attacks from noisy trajectories. This paper experimentally evaluates such a game-theoretic framework on a mobile robot. Although the framework predicts constant attacks and likelihood-ratio tests as equilibrium strategies, robot experiments show large errors in the predicted detection failure rate due to a mismatch in final-state variance. We model this effect using empirical Gaussian final-state distributions. Experiments and simulations reduce the prediction error to 5% and clarify the model's limits.

Keywords: Smart structures and vibration control, Mechatronic system estimation, identification, control, Mechatronics for advanced manufacturing and energy systems
Abstract: Conventional Electromagnetic Vibration Energy Harvesters (EVEHs) can be inefficient when tuned to fixed impedances, particularly under the non-stationary, broadband conditions typical of railway environments. To overcome this limitation, this paper introduces an adaptive impedance matching controller driven by a Machine Learning (ML) surrogate model. By leveraging a Random Forest (RF) regression trained on statistical signal features, the proposed system predicts the optimal complex load impedance in real-time, enabling precise complex conjugate matching. Experimental validation confirms that the controller not only tracks the theoretical maximum power during sinusoidal sweeps but also significantly outperforms traditional fixed-tuning strategies in real-world benchmarks. Specifically, under non-stationary railway vibration profiles, with instantaneous power improvements exceeding 20% during off-resonance events, proving it is the most reliable power source for automated condition monitoring systems.

Keywords: Task and motion planning, Aerial, field, and marine robotics
Abstract: This paper presents a scalable framework for multi-robot task allocation in complex environments where estimating task execution costs is computationally expensive. While combinatorial auction-based approaches offer reliable solutions, the exponential complexity of bundle generation typically renders them intractable for real-time reactive applications, particularly when accurate path planning is required for cost validation. We address this through a distributed, two-stage multi-fidelity bundle generation approach. Agents utilize a local search tree guided by a low-fidelity heuristic (such as euclidean distance) to rapidly explore the bundle space, applying high-fidelity path planning only to the most promising candidates in a best-first manner. These refined bids are then submitted to a central coordinator that solves a set packing problem to ensure global feasibility and maximize the overall utility. Simulation results in multiple environments demonstrate that the framework is able to improve the performance of reactive auction-based task allocation. Overall, the presented framework is shown to enable reactive task allocation with dynamic bundle sizes in multiple settings without exposing the agents' state and internal cost estimation models.

Keywords: Task and motion planning, Autonomous navigation
Abstract: This paper presents a framework for multi-mission multi-robot task allocation that integrates continuous-time routing with a lightweight bucketed reservation layer. Rather than collapsing all objectives into a single global mission, the framework keeps missions distinct and enables controlled sharing of robots across stakeholders with differing priorities and limited information exchange. The reservation layer overlays coarse time buckets on the planning horizon, allowing planners to specify time-phased mission quotas and enforce one-mission-per-robot commitments within each interval, all while preserving continuous task timing. This structure provides an operational control interface through which operators can adjust mission priorities over time without disclosing internal task details, enabling responsive, interpretable, and privacy-aware coordination. The results show that the proposed framework delivers feasible, continuous-time schedules that respect mission-level policies and achieve coordinated mission progress.

Keywords: Task and motion planning, Autonomous navigation, High-performance motion control systems
Abstract: pRRTC is a GPU-accelerated RRT-Connect algorithm that uses parallelism for both sample expansion and collision detection. However, setting the sample size for discrete collision detection equal to the number of threads per block may not meet the finer collision resolution required by certain applications. In this paper, we leverage a lazy strategy to enhance the efficiency of pRRTC to mitigate the safety and discretization accuracy tradeoff set by default number of threads per block. Our approach reduces a significant number of fine collision detection by deferring fine full path collision check until after the initial path linking start and goal is generated by pRRTC with its default number of discretization. Simulations in environments with 35 randomly placed rectangular obstacles and walls with narrow passages show that in safety-aware fine discretization lazy-pRRTC achieves accurate tree extension with approximately 3× higher efficiency than its predecessor, pRRTC, and enables efficient waypoints generation for fast navigation in harder environments due to significantly fewer state collision checks.

Keywords: Task and motion planning, Autonomous navigation, Robotic learning and adaptation
Abstract: This paper introduces a certification framework that analyzes deep reinforcement learning policies used in autonomous surgical planning. Current learning-based controllers lack formal safety guarantees, and we address this by representing Deep Reinforcement Learning (DRL)-generated surgical plans as explicit Markov decision processes. First, the feasibility of a surgical plan is established by two conditions: absorption stability at the target state and finite-time reachability to it. After the feasibility assessment, a quantitative robustness index is derived from a reachability-layer decomposition. This index measures the resilience of the surgical plan when a single state transition is disrupted such as by tissue deformation. Finally, the theoretical approach has been implemented in an interactive visual interface for verification and evaluation. The effectiveness of this framework has been verified through an illustrative simulation on an ultrasound navigation task and identify the critical transitions required to reach the target position.

Keywords: Task and motion planning, Autonomous navigation, Robotic learning and adaptation
Abstract: Bio inspired metaheuristic algorithms are optimization methods that mimic natural phenomena, biological evolution to solve complex problems. This paper proposes a hybrid navigation framework combining Teaching-Learning-Based optimization(TLBO) algorithm for Bézier curve path planning and Geometric Model Predictive Control (GMPC) for trajectory tracking in a dynamic environment, implemented on a skid-steered mobile robot. Experimental validation across 45 trials with varying obstacle configurations and human interaction scenarios demonstrates framework accuracy of 79.8%±2.1% in simulation and 70.7%±21.2% accuracy in real-time experiment with significant performance observed in dynamic human interaction scenarios.

Keywords: Task and motion planning, Mechatronic system modeling, design, optimization, High-performance motion control systems
Abstract: Underactuated overhead cranes present significant challenges in trajectory planning due to their complex nonlinear dynamics, coupling effects between actuated and unactuated states, and the necessity of real-time feasibility. To bridge the gap between theoretical research and industrial application, this paper proposes a full-state constrained trajectory planning framework that ensures dynamic feasibility while maintaining real-time computational performance. The proposed method explicitly incorporates system dynamics and full-state constraints into the optimization process, enabling simultaneous regulation of both actuated and unactuated variables. A partial model simplification strategy is introduced to accelerate computation without sacrificing dynamic consistency, allowing real-time online trajectory generation. The framework also demonstrates robustness against modeling uncertainties and effectively balances multiple objectives, including obstacle avoidance, motion smoothness, and time efficiency. Extensive simulations and experimental validations on overhead crane systems verify the framework’s effectiveness, achieving dynamically feasible and smooth trajectories with precise control of unactuated variables under complex operating conditions.

Keywords: Task and motion planning, Robot perception and sensing, Mechatronic system modeling, design, optimization
Abstract: Automatic visual quality inspection is a critical application in modern manufacturing, leveraging robotics and computer vision to improve efficiency and precision. Previous methodologies often address the inspection challenge from a singular perspective of robotics or computer vision, which constrains the performance and generalization of the inspection performance.

This work presents a new framework focused on refining the inspection pose (viewpoints) candidates to improve the overall coverage. This process integrates the sensor model, environment constraints for collision avoidance, the kinematics of the robotic system, and the model of the inspected object.

The final inspection plan is computed by adopting a path planner to derive a collision-free trajectory and visit the viewpoints in the order obtained by solving the Traveling Salesman Problem.

Our framework is extensively evaluated in simulation and compared to the state of the art, demonstrating superior performance in achieving extensive coverage. Real-world experiments are conducted to prove the effectiveness of our methods. In both cases, results are presented for different objects and two robotic setups: (i) a robot with 6-dof and (ii) a robotic system with 7-dof.

Keywords: Task and motion planning, Robotic grasping and manipulation
Abstract: This paper presents a nonlinear Model Predictive Control (MPC) planner for dynamic object interception using cooperative manipulator systems under closed-chain constraints. We introduce an Adaptive-Terminal (AT) formulation that employs cost shaping to mitigate actuator power violations common in Primitive-Terminal (PT) approaches. Experimental validation on a physical platform demonstrates superior motion quality and robustness compared to the PT baseline. Crucially, the system exhibits excellent real-time performance, achieving an average computation time of 19ms -- less than half the 40 ms sampling interval. This establishes the framework's suitability for agile, safety-critical cooperative tasks.

Keywords: Task and motion planning, Robotic grasping and manipulation, AI-powered robotics
Abstract: Voxel-grid reinforcement learning is commonly used for path planning in redundant manipulators due to its simplicity and reproducibility. However, direct execution through point-wise numerical inverse kinematics on 7-DoF arms often yields step-size jitter, abrupt joint transitions, and instability near singular configurations. This work proposes an offline bridging framework that enables smooth continuous execution without modifying the core discrete planning architecture. On the planning side, step-normalized 26-neighbour Cartesian actions with geometric tie-breaking reduce unnecessary turns and oscillations. On the execution side, a task-priority damped least-squares (TP-DLS) inverse kinematics (IK) solver ensures stable tracking through null-space posture regulation and joint centering under trust-region and velocity constraints. Experiments on a 7-DoF manipulator show that this bridge improves planning success in dense scenes from 0.58 to 1.00, shortens representative path length from 1.53 m to 1.10 m, and reduces peak joint accelerations by over an order of magnitude while maintaining sub-millimeter end-effector accuracy. These results demonstrate that discretely planned RL paths can be made reliably executable through principled integration with established IK techniques.

Keywords: Time/parameter varying system identification, Linear system identification
Abstract: This paper introduces a non-iterative parametric identification algorithm for linear time-periodic (LTP) systems. The proposed method reduces the identification task to solving a set of linear equations and yields a state-space representation in the observable canonical form. This frequency-domain approach leverages periodic input test signals and enables model complexity reduction through truncation of the harmonic transfer functions. The resulting approach provides an efficient and structured framework for modeling LTP systems.

Keywords: Time/parameter varying system identification, Linear system identification
Abstract: This paper proposes a recursive algorithm, rARX-DIPCA, for identifying errors-in-variables autoregressive models with exogenous input (EIV-ARX), for tracking time-varying SISO processes. Building on a recently developed recursive iterative PCA method, the proposed algorithm recursively updates model parameters and noise variances as new measurements arrive, without storing historical data beyond a specified lag window. The method enables real-time adaptation to sensor degradation, and changes in model coefficients. The algorithm simultaneously identifies process order, time delay, and noise variances while maintaining computational efficiency through online covariance updates. Simulation studies on benchmark systems demonstrate effective tracking performance and practical applicability.

Keywords: Time/parameter varying system identification, Linear system identification, Data-driven control theory
Abstract: This paper presents a data-driven framework for identifying linear discrete-time periodic (LDTP) systems and extracting their Floquet-equivalent models. Identification of LDTP systems is challenging due to periodically varying state-transition matrices, while Floquet reduction requires numerically sensitive matrix-root computations of the monodromy matrix. The proposed approach integrates an optimization-based estimator with a numerically robust Schur–Pad´e procedure for computing the principal P-th matrix root of the monodromy. A Monte Carlo study on randomly generated stable systems examines how system order, period length, and monodromy conditioning affect both identification accuracy and Floquet feasibility. The resulting workflow provides a reliable and systematic route for recovering periodic dynamics and their Floquet structure using only input–output data

Keywords: Time/parameter varying system identification, Nonlinear system identification, Machine and deep learning for system identification
Abstract: Identifying control-friendly models of nonlinear systems remains one of the major challenges at the intersection of system identification and control. The Linear Parameter-Varying (LPV) framework offers a promising solution, but existing identification methods often rely on model structures with affine scheduling dependency. Instead, this work proposes the use of LPV models with Linear Fractional Representation (LFR) admitting a rational scheduling-dependency, capable of modelling complex nonlinear systems with fewer scheduling variables compared to affine models. This work introduces a direct parameterization to ensure well-posedness of rational LPV-LFR models, which by joint-estimation of an LPV plant and scheduling map, using only input-output data, is capable of modelling complex nonlinear systems. Accuracy of the proposed approach is shown on two simulation examples.

Keywords: Wearable robotics, Human-robot interaction, Humanoid and legged robots
Abstract: This work introduces RoboWear21, an asymmetric lower-limb exoskeleton developed to accommodate the differing mechanical demands of paretic and non-paretic limbs. The device integrates side-specific actuators, passive hip DOFs, and a hierarchical controller combining gravity compensation, disturbance observer, and gait-phase-dependent torque generation. Gait state is estimated through an IMU-based swing detection scheme and an adaptive oscillator that maps hip motion to a continuous phase variable. Bench and user tests with three healthy participants demonstrated joint tracking RMSE up to 2.177°, phase estimation with an overall RMSE of 1.191 ± 0.894% (R² = 0.997 ± 0.002), and gravity-compensation deviations within 0.022°, suggesting the system's suitability for individualized assistance in hemiparetic gait rehabilitation.

Keywords: Wearable robotics, Medical and rehabilitation robotics, Human-robot interaction
Abstract: Hip-assistive wearable robots are lightweight, portable, and easy to use, but they typically lack foot-mounted sensors, making accurate identification of gait events particularly challenging in asymmetric or pathological gait. Existing approaches have either relied on additional shoe sensors or have been validated only on healthy users, limiting their applicability in sensor-minimal configurations and abnormal walking conditions. This study proposes an LSTM–based stance and swing state recognition framework using only absolute thigh angle signals obtained from a hip-assistive wearable robot. In the implemented bilateral configuration, left and right thigh-angle sequences are processed by limb-specific LSTM encoders and fused to predict stance and swing states for both limbs. Experiments on normal walking and hemiplegic-like asymmetric gait achieved approximately 87% accuracy without using foot sensors as model inputs. The full estimation pipeline was further implemented in a pseudo-online and real-time setting, demonstrating its feasibility for embedded execution on wearable robots.

Keywords: Autonomous marine systems and vehicles, Marine robotics, Perception and filtering in marine systems
Abstract: Underwater robotic perception degrades with colour loss, reduced contrast, and backscatter, yet it remains unclear which cues best signal when performance will hold. We analyse matched in-air/underwater data across lighting, range, and scene complexity, relating condition-wise appearance changes to downstream robustness. A consistent pattern emerges: preserved red–green colour balance is the strongest positive indicator of reliability, while losses of green-band detail and brightness are the most reliable warnings of failure. These task-aware indicators explain most variability across conditions and provide a compact signal for onboard health monitoring, adaptive planning, and mission triage in challenging underwater environments.

Keywords: Optimal control theory, Applications of optimal control, Differential or dynamic games
Abstract: This paper investigates the energy management problem for large populations of electric vehicles (EVs) with finite-continuous hybrid states. First, a novel model is proposed that integrates continuous state of charge and discrete events triggered by on-off charging mode switches. Then, a hierarchical optimization framework is developed to cope with the hybrid system. In this framework, the upper level, managed by the grid operator, achieves macroscopic load balancing for large populations of EVs by optimizing the finite state transition rates; the lower level involves decentralized decision-making, where individual EVs adjust their charging power to optimize their respective objectives. Given the analytical challenges posed by large-scale EVs charging behaviors, this paper formulates the coordination problem of the EV population as a mean-field game (MFG), where its equilibrium solution is characterized by two coupled sets of Hamilton-Jacobi-Bellman (HJB) and Fokker-Planck (FP) equations. Compared to conventional HJB-FP equations, these equations incorporate additional terms associated with the finite state transition behavior. Furthermore, theoretical analysis shows that the MFG provides an varepsilon-Nash equilibrium for a finite number of EVs. Finally, an efficient numerical solution is illustrated for the optimal control problem, and simulation results demonstrate the effectiveness of the proposed framework and methodology.

Keywords: Optimal control theory, Applications of optimal control, Numerical methods for optimal control
Abstract: Nitrogen-vacancy (NV) centers are promising experimental platforms for quantum information processing. In this paper, we investigate their controllability and fundamental quantum speed limit for two-qubit gates. Such a quantum system consists of two coupled spins, an electronic and a nuclear spin, where only the former can be controlled directly via microwave pulses. We discuss the various physical approximations that lead to the system model before studying its controllability. We use this control issue as an example to demonstrate how standard geometric control tools can be applied to spin networks. We complete this analysis with a computation of the quantum speed limit using known analytical techniques on Lie groups and their algebras. We finally demonstrate, thanks to preliminary optimal control numerical experiments, that this limit can be approached while keeping a reasonable energy of the control field.

Keywords: Optimal control theory, Linear systems
Abstract: Discrete-time minimum ell_1-norm has often been suggested as a solution for sparse optimal control of linear time-invariant systems. Unlike the continuous-time case, where controllability is guaranteed to provide the sparsest solution, this is no longer true in discrete-time. We propose a deterministic understanding of failure cases, leveraging the framework of total positivity to derive conditions under which the sparsest solution cannot be recovered. Thus, providing insights into the a priori design of sparse optimal control problems, as well as some more general compressed sensing settings, explaining why such failure is to be predicted.

Keywords: Optimal control theory, Optimal control of PDE systems, Control of distributed parameter systems
Abstract: This work focuses on indirect descent methods for optimal control problems governed by nonlinear ordinary differential equations in Banach spaces, viewed as abstract models of distributed dynamics. As a reference line, we revisit the classical schemes, rooted in Pontryagin’s maximum principle, and highlight their sensitivity to local convexity and line-search procedures. We then develop an alternative method based on exact cost-increment formulas and finite-difference probes of the terminal cost. Numerical results for an Amari-type neural field illustrate monotone decrease of the cost, obtained without solving the adjoint equation.

Keywords: Optimal control theory, Stability of nonlinear systems, Lagrangian and Hamiltonian systems
Abstract: This paper revisits the infinite-horizon optimal control (IOC) problem from the perspective of a family of extremals emanating from the local stable manifold of the associated Hamiltonian system. We analyze conditions under which these extremals—parameterized by their root points on the manifold—form a Lagrangian submanifold, thereby yielding a stabilizing solution to the Hamilton–Jacobi–Bellman equation (HJBE).

We further investigate how the emergence of conjugate points—instances where the Riccati differential equation along an extremal blows up—destroys this geometric structure. Additionally, we explore the connection between conjugate points and the uniqueness of solutions to a two-point boundary value problem (BVP) that incorporates the local stable manifold as a terminal condition. This BVP facilitates the generation of neighboring extremals around a reference extremal.

Numerical examples using a cart-inverted-pendulum system illustrate these geometric properties through families of extremals corresponding to swing-up maneuvers and extremals exhibiting conjugate points that break the embedded submanifold structure.

Keywords: Optimal control theory, Stochastic optimal control problems, Numerical methods for optimal control
Abstract: We consider a general aggregation framework for discounted finite-state infinite horizon dynamic programming (DP) problems. It defines an aggregate problem whose optimal cost function can be obtained off-line by exact DP and then used as a terminal cost approximation for an on-line reinforcement learning (RL) scheme. We derive a bound on the error between the optimal cost functions of the aggregate problem and the original problem. This bound was first derived by Tsitsiklis and van Roy [TvR96] for the special case of hard aggregation. Our bound is similar but applies far more broadly, including to soft aggregation and feature-based aggregation schemes.

Keywords: Optimization-based estimation and control, Design methods for data-based control, Real-time optimal control
Abstract: Feedback optimization steers the steady-state operation of dynamical systems to optimal operating points. However, most existing methods still require exact knowledge of the plant dynamics, which is rarely available in practice. In this paper, we introduce a randomized two-point gradient-free feedback optimization method inspired by zeroth-order optimization. Our controller evaluates plant performance at two points to estimate gradients and update control inputs in real-time. For problems with smooth, nonconvex objectives, our method achieves convergence to an ε-stationary point with iteration complexity O(pε-1), where p denotes the dimension of the input vector, thereby recovering the best-known bounds for static two-point optimization. Numerical simulations support the theoretical results.

Keywords: Optimization-based estimation and control, Nonlinear control of switched & hybrid systems, Control of hybrid systems
Abstract: This paper proposes a new command governor (CG) scheme for the tracking of discrete-time switched linear systems subject to input and state constraints. The approach leverages a novel class of admissible sets, termed switch-dependent semi-ellipsoidal admissible sets, which exploit available information on the switching signal. These sets enable the design of a recursively feasible CG that guarantees closed-loop constraint satisfaction. The proposed approach is demonstrated through a numerical example.

Keywords: Optimization-based estimation and control, Observer design, Linear systems
Abstract: Sensor placement for maximizing the estimation performance of the Kalman filter is an NP-hard optimization problem. Furthermore, its feasible set grows combinatorially with the candidate locations and the number of sensors. In this paper, we study this sensor placement problem for a 3D thermoelastic system modelled as a discrete-time linear stochastic model. We use the Novel Binary Artificial Bee Colony (NBABC) algorithm with a Gramian-based pre-filter to reduce the computational complexity. Our results show the efficiency and the fast convergence of the proposed approach.

Keywords: Parametric optimization, Convex optimization, Large-scale and networked optimization problems
Abstract: We introduce a principled learning to optimize (L2O) framework for solving fixed-point problems involving general nonexpansive mappings. Our idea is to deliberately inject summable perturbations into a standard Krasnosel'skii–Mann iteration to improve its average-case performance over a specific distribution of problems while retaining its convergence guarantees. Under a metric sub-regularity assumption, we prove that the proposed parametrization includes only iterations that locally achieve linear convergence—up to a vanishing bias term—and that it encompasses all iterations that do so at a sufficiently fast rate. We then demonstrate how our framework can be used to augment several widely-used operator splitting methods to accelerate the solution of structured monotone inclusion problems, and validate our approach on a best approximation problem using an L2O-augmented Douglas–Rachford splitting algorithm.

Keywords: Parametric optimization, Design methods for data-based control, Optimization-based estimation and control
Abstract: High-precision positioning in manufacturing equipment requires fast settling with minimal vibration. The asymmetric S-curve (AS-curve) is a jerk-limited trajectory that enables high speed and precision, but its many tuning parameters make adjustment difficult. Bayesian Optimization (BO) is a well-established sample-efficient optimization method, but its performance can be improved by exploiting information closely related to control performance. We propose waveform-aware BO for sample-efficient AS-curve shaping. A Gaussian process regression (GPR) incorporating a distance metric between command waveforms yields an accurate model with few evaluations and accelerates BO convergence. Experimental results on a real-world setup demonstrate equivalent tuning using only 15% of the trials required by conventional BO.

Keywords: Parametric optimization, Model reduction of distributed parameter systems
Abstract: This paper addresses the problem of model reduction for parameter dependent systems where the reduction criterion expresses a design objective for the parameter dependent system. Specifically, we develop a reduction method for systems that are required to meet an explicit guarantee on the H₂ approximation error with respect to a design objective. This guarantee is combined with efficiency improvements on the reduction scheme and an error estimation. The performance of the method is illustrated on a thermal design optimization problem. Results indicate superior computational efficiency compared to classical methods.

Keywords: Randomized algorithms in robust control, Distributed parameters port Hamiltonian systems, Robust estimation
Abstract: This work addresses the problem of distributed online estimation in a dynamic and potentially heavy-tailed environment. The proposed distributed stochastic approximation algorithm incorporates momentum and operates under H¨older smoothness, Lyapunov strong convexity, functional drift, and sparse structural shifts. Synthetic tests on a drifting multi- dimensional Rosenbrock function with heavy-tailed noise confirm bounded tracking error and rapid recovery following abrupt changes. Equity market experiments further validate the method, yielding stable estimates for portfolio risk management and intraday mean-reversion strategies.

Keywords: Real-time optimal control, Control barrier functions and state space constraints, Adaptive control design
Abstract: This paper addresses the motion control problem for mobile robots in obstacle-cluttered environments. The mobile robot has partial environment information only, and aims to move from an initial position to a target position without collisions. For this purpose, a reactive planning based control strategy (RPCS) is proposed. First, the initial and target positions are connected as a reference trajectory. Then, a reactive planning strategy (RPS) is developed to ensure the collision avoidance by modifying the reference trajectory locally based on the partial environment information. Next, an adaptive tracking control strategy (ATCS) is proposed to track the reference trajectory with potentially local modifications via the discretization techniques. Finally, the RPS and ATCS are combined to establish the RPCS, whose efficacy and advantages are illustrated by numerical examples.

Keywords: Real-time optimal control, Optimal control theory, Convex optimization
Abstract: This paper proposes an intrinsic pseudospectral convexification framework for optimal control problems with manifold constraints. While pseudospectral successive convexification combines spectral collocation with successive convexification, classical pseudospectral methods are not geometry-consistent on manifolds. This is because interpolation and differentiation are performed in Euclidean coordinates. We introduce a geometry-consistent transcription that enables pseudospectral collocation without imposing manifold constraints extrinsically. The resulting method solves nonconvex manifold-constrained problems through a sequence of convex subproblems. A six-degree-of-freedom landing guidance example with unit quaternions and unit direction vectors demonstrates the practicality of the approach. The proposed method preserves manifold feasibility to machine precision and achieves significant computational speedups.

Keywords: Robust controller synthesis, Analytic design, Linear systems
Abstract: This paper presents a stable plug-in controller design that improves the closed-loop performance of pre-stabilized single-input-single-output (SISO) linear time-invariant (LTI) systems without sacrificing inherent robustness. To ensure both controller stability and desired pole placement, the problem is reformulated via an s-domain transformation psi(s) = s - alpha (alpha > 0). This shifts the stability boundary, rendering the original system virtually unstable and converting the design into a strong stabilization problem. By analytically solving the transformed system and applying an inverse shift, the proposed non-iterative approach yields low-order controllers. Experimental validation on a magnetic levitation system demonstrates significantly improved tracking and leftward-shifted poles compared to a standalone proportional-integral-derivative controller.

Keywords: Robust controller synthesis, Switching stability and control
Abstract: We introduce a second-order version of the hybrid integrator-gain system (HIGS). In the proportional mode the second state is either reset to zero or tracks the input. We derive a method for computing the describing function and higher-order harmonics in terms of a matrix exponential. In comparison to the HIGS the new element shows the amplitude response of a second order system whereas the phase drops to approximately −52°. Using a sector transformation we can show that the second-order HIGS is passive, which allows for a conservative circle-criterion-like condition to test for closed-loop stability.

Keywords: Robust estimation, Learning methods for optimal control, Linear time-delay systems
Abstract: This paper presents a generalized repetitive-control (GRC) framework that achieves both precise tracking of periodic signals and suppression of aperiodic disturbances. The relationship between the ideal periodic internal model and the GRC structure is analyzed. Based on this analysis, a second-order Butterworth filter and a time-delay parameter are designed to ensure accurate steady-state tracking. In addition, the inherent limitation of the conventional equivalent-input-disturbance (EID) estimator is identified. The conventional EID estimator behaves as an integrator and therefore responds slowly to disturbances. To overcome this problem, a proportional-integral EID (PI-EID) estimator is developed. The new PI-EID estimator provides fast disturbance compensation while maintaining high estimation accuracy. The stability of the control system is guaranteed. Simulation results demonstrate that the proposed method significantly improves steady-state tracking accuracy when compared with modified repetitive control and complex-coefficient-filter-based repetitive control. The proposed method also achieves superior transient and steady-state disturbance rejection when compared with the conventional EID method and the improved EID method.

Keywords: Robust time-delay systems, Decentralized control, Analytic design
Abstract: This paper addresses the consensus problem for linear multi-agent systems with heterogeneous time-varying communication delays. Existing delay-dependent approaches based on Lyapunov--Krasovskii functionals and LMIs often suffer from high computational complexity and limited analytical insight. To overcome these limitations, an explicit modal decomposition framework is developed that exploits the Laplacian eigenstructure to decouple the network dynamics into independent subsystems. For each mode, delay-dependent stability conditions are derived in closed algebraic form using Sylvester’s criterion, enabling direct characterization of admissible delays and controller gains without numerical optimization. For agents with arbitrary relative degree, a dynamic high-gain controller is introduced to ensure simultaneous stabilization of all nonzero Laplacian modes under slowly varying heterogeneous delays. The proposed approach provides scalable and analytically tractable stability conditions that explicitly reveal the influence of network topology on delay robustness. Numerical examples demonstrate convergence to consensus and bounded control effort under time-varying delays.

Keywords: Robust time-delay systems, Robust controller synthesis, Control of hybrid systems
Abstract: This paper develops a recursive solution to the state-feedback control problem for linear discrete-time systems with unknown random state delays and norm-bounded parametric uncertainties. It is assumed that the rate of variation between consecutive delays is bounded, and an unobserved Markov chain is used to model stochastic delay behavior. By employing the lifting technique, the original state-delayed system is converted into an equivalent delay-free Markovian jump linear system formulation. Leveraging this framework, an optimization problem is formulated that accounts for the impact of delayed state while simultaneously accommodating worst-case uncertainties. The stabilizing gains are then obtained via recursive Riccati equations, which establish standard conditions for stability and convergence. The performance of the proposed robust regulator is illustrated using a model of an F-16 aircraft. We present a comparative study using robust H_{infty} state-feedback controllers to demonstrate the effectiveness of the developed recursive regulator.

Keywords: Robustness analysis, Controller constraints and structure, Stability of nonlinear systems
Abstract: For a nonlinear system in Byrnes-Isidori form, subject to unbounded perturbations, i.e. perturbationsthat satisfy a given bound only locally on a subset of the state space, we apply the continuous approximation of Lyapunov redesign within the feedback linearisation model-following control (MFC) scheme for trajectory tracking. We establish practical tracking by generalising a tube-based stability analysis proposed for single-loop control to MFC. Conceptually, we exploit that the Lyapunov function used for the Lyapunov redesign satisfies a differential inequality, thereby guaranteeing that the solution of the perturbed closed loop remains in a tube along the a-priori known solution of the model simulated in the model control loop.

Keywords: Robustness analysis, Robust controller synthesis, Robust linear matrix inequalities
Abstract: This paper develops a linear matrix inequality (LMI) condition for the parametric quadratic stabilizability of bimodal piecewise linear systems under affine state feedback. The affine reference input induces equilibrium migration across switching regions. The proposed condition guarantees the existence and uniqueness of the equilibrium point together with quadratic stability of the closed-loop system.

Keywords: Robustness analysis, Uncertain systems, Linear systems
Abstract: Robustness analysis of uncertain nonlinear systems is often dominated by computationally expensive Monte-Carlo simulations, motivating the development of alternative approaches, including deterministic methods for worst-case assessment. An efficient solution approach is developed for a finite-horizon robustness analysis method that is based on a linear time-varying model along a nominal trajectory with quadratic constraints capturing nonlinear effects. The method leverages a transformed Riccati differential equation formulation with analytically optimized time-varying parameters to reduce computational complexity. Local quadratic constraints are iteratively refined using sparse grids. Application to Huygens' atmospheric entry flight demonstrates accurate estimation of worst-case bounds with moderate conservatism.

Keywords: Stochastic optimal control problems, Learning methods for optimal control, Optimal control theory
Abstract: This paper investigates the static output feedback control problem for linear quadratic regulation (LQR) in discrete-time stochastic systems with state- and controldependent noises. To solve the stochastic LQR problem, policy iteration (PI) and value iteration (VI) algorithms are provided. Furthermore, via the provided intermediate matrix technique, a comparative analysis of the convergence rates for the given PI and VI algorithms is presented, along with a detailed proof. Finally, simulation examples of the F-16 aircraft model are conducted to verify the effectiveness of the proposed algorithms and the validity of the relevant theories.

Keywords: Application of nonlinear analysis and design
Abstract: A design procedure of active vibration absorbers for underactuated systems is presented. The control design is based on the exact feedback linearization of the system, with the primary structure being the linearizable part. The paper focuses on two systems: the 4-link with two actuators and the Acrobot with two vibration absorbers. The results are demonstrated using two examples.

Keywords: Applications of optimal control, Numerical methods for optimal control, Stochastic optimal control problems
Abstract: Multi-scale thermal environment control requires a fast control framework that explicitly accounts for uncertainties such as disturbances. In this study, we propose a hierarchical covariance steering (CS) method based on two spatiotemporal resolutions. First, we demonstrate the effectiveness of CS for a heat diffusion system through an indoor temperature field control problem. Then, we organize the problem formulation and theoretical foundation of hierarchical CS and discuss its effectiveness.

Keywords: Batch and semi-batch process control, Process modeling, identification, and estimation techniques, Real-time optimization and control in chemical processes
Abstract: Optimal control growth in industrial sugar production requires precise supersaturation control in the metastable region. Due to limited online measurement capability, we utilized a model-driven soft sensor for real-time estimation. A feedforward-feedback control architecture was developed to regulate the process within optimal limits. The proposed strategy enables reliable control and easy deployment in industrial environments.

Keywords: Control architectures in automotive control, Electric and solar vehicles, Nonlinear and optimal automotive control
Abstract: Thermal management systems (TMS) of battery electric vehicles (BEVs) represent a key enabler for further energy savings. Such systems enhance efficiency by recovering waste heat from key powertrain components as the electric drive, power electronics, and HV battery and by employing an integrated heat pump for active thermal conditioning of the vehicle. This high complexity in system design and the growing number of vehicle variants requires a generalizable control software architecture to serve industrial projects. Concurrently, many scientific researchers are actively working on TMS and propose promising predictive and advanced control designs. This discussion paper will explain the corresponding challenges and give some best-practice solutions for bridging the gap.

Keywords: Control engineering curricula, Control education laboratories
Abstract: Climate change and population growth pose new challenges for water management. Control theory provides additional tools to meet these challenges. To provide future water managers with insight into those tools, a course that looks at control theory from a water management perspective is essential. Examples for such a course should ideally be closely related to real water systems. To make possible more realistic examples, a toolkit is under development that allows students to use Python as a user interface to create and run hydrodynamic models in Delft3D FM, to analyze the run results from Python, and to control weirs, gates, and pumps in the model with Python code.

Keywords: Data fusion and mining in control, Digital twins for cyber physical systems, Knowledge-based and data-driven control
Abstract: In data processing chains, different types of data can be transformed into unified dimensionless indicators which include efficiently nonlinear effects of the measurements. Informative indicators, which are based on generalised norms and nonlinear scaling are the elements of the advanced deep learning. Different types of data, analysis methodologies, system structures and interactions follow the phases of cyber-physical systems and provide a feasible basis for configuration. Several small-scale models support development of cyber-physical systems (CPS). Each module can be a model, diagnostic or control unit.

Keywords: Data-driven control theory, Time series modeling, Learning methods for control
Abstract: This paper proposes a novel data-driven sub-optimal servomechanism design from noisy response data. As widely known, servomechanism controllers incorporate state feedback and integral action to achieve robust tracking without steady-state error. In line with the data informativity approach, we derive a linear matrix inequality condition for a data-driven sub-optimal servomechanism design from noisy input-state-output data taking account of the internal model structure. The resulting servomechanism controller guarantees the integral quadratic performance below a prescribed level, improving transient response. The effectiveness of the proposed method is demonstrated through a numerical example.

Keywords: Data-driven control theory
Abstract: Data-driven control of nonlinear systems with rigorous guarantees is a challenging control problem. Integral quadratic constraints (IQCs) provide a powerful framework for modeling nonlinearities. This paper presents a data-driven min-max model predictive control (MPC) synthesis method for unknown systems subject to (nonlinear) uncertainties using the IQC framework. The unknown system matrices are characterized by a set-membership representation using the input-state data and the knowledge of the IQCs. We derive two semidefinite programs (SDPs) that minimize an upper bound on the worst-case cost over all possible system dynamics and uncertainties. By iteratively solving these SDPs, the proposed state-feedback control law is obtained. We further prove that the resulting closed-loop system is exponentially stable and satisfies the input and state constraints. A numerical example demonstrates the validity of the proposed method.

Keywords: Machine and deep learning for system identification, Learning methods for control, Data-driven control theory
Abstract: Aerial manipulators undergo rapid, configuration-dependent changes in inertial coupling forces and aerodynamic forces, making accurate dynamics modeling a core challenge for reliable control. Analytical models lose fidelity under these nonlinear and nonstationary effects, while standard data-driven methods such as deep neural networks and Gaussian processes cannot represent the diverse residual behaviors that arise across different operating conditions. We propose a regime-conditioned diffusion framework that models the full distribution of residual forces using a conditional diffusion process and a lightweight temporal encoder. The encoder extracts a compact summary of recent motion and configuration, enabling consistent residual predictions even through abrupt transitions or unseen payloads. When combined with an adaptive controller, the framework enables dynamics uncertainty compensation and yields markedly improved tracking accuracy in real-world tests.

Keywords: Data-driven control theory, Learning methods for control, Machine and deep learning for system identification
Abstract: Inverse optimal control provides a principled framework for identifying the cost function of an optimal control problem (OCP) from observed state trajectories, assuming known system dynamics. This work addresses infinite-horizon OCPs for linear and nonlinear systems and proposes a neural network-based algorithm that simultaneously reconstructs the underlying cost and a stabilizing feedback law from pure state data. The approach integrates data fidelity with stability constraints to ensure physically meaningful solutions. Numerical experiments demonstrate the effectiveness of the method for different systems and under varying noise levels.

Keywords: Data-driven methods for FDI/FTC, AI methods for FDI/FTC, Fault detection and isolation methods
Abstract: The integration of model-based and data-driven paradigms provides a powerful framework for fault diagnosis by combining the interpretability of analytical redundancy relations, i.e., input-output relations that are used as diagnosis indicators in model-based diagnosis, with the adaptability of learning techniques. DT4X is a recent diagnosis algorithm that uses symbolic regression to generate multivariate relations leveraging some properties of analytical redundancy relations and uses them as split functions in a decision tree. However, its symbolic regression procedure optimizes only the separation between two selected classes at each node, often fragmenting the remaining classes and degrading both interpretability and diagnosis performance. This paper introduces DT4X+, an enhanced version of DT4X that modifies the construction of training sets and the symbolic-regression loss so that expressions separate the target classes while preserving the coherence of non-target classes. The resulting relations become fully consistent with ARR properties and lead to more informative splits, improved robustness, and better performance on dynamic-system datasets. Experiments conducted on several benchmark systems demonstrate the benefits of this enhanced formulation.

Keywords: Data-driven control theory, Randomized algorithms in stochastic systems
Abstract: We study the problem of optimizing the stationary behavior of a discrete-time linear system subject to a stochastic disturbance. Specifically, our objective is to design a disturbance compensator that shapes the stationary state and control input distribution so as to guarantee a certain performance subject to joint state-input constraints. We consider the case when a dataset of finite-length disturbance realizations is available for the compensator design, and propose a data-driven approach that integrates scenario optimization with a mechanism to reduce the stationary state to a moving average process of finite order. The proposed method is shown to outperform a competitive state-of-the-art method.

Keywords: Distributed optimization, Distributed control and estimation
Abstract: This paper presents a real-time computational framework for multi-node distributed optimization by extending the Augmented Lagrangian Alternating Direction Inexact Newton (ALADIN) algorithm. Our approach integrates adjoint sequential quadratic programming (SQP) techniques to enable efficient approximation of Jacobian information within the ALADIN embedded quadratic program, thereby reducing communication overhead. Furthermore, to decrease computational complexity, we design an event-driven update strategy that avoids updating Hessian and Jacobian matrices at every iteration. The proposed method maintains convergence guarantees while achieving a twofold improvement in computational speed, making it suitable for time-sensitive applications with strict timing constraints. Numerical experiments demonstrate that our approach achieves competitive performance while exhibiting superior computational efficiency in real-time scenarios, validating its practical applicability for time-sensitive distributed optimization challenges.

Keywords: Distributed optimization, Multi-agent systems
Abstract: There is no general method for designing proper parameters to achieve faster convergence in distributed optimization algorithms. In this paper, we consider the distributed inexact gradient tracking (DIGing) algorithm with the objective function being the unweighted sum of squares. By representing the iteration algorithm as a dynamical linear system, we decompose it into different graph frequencies and obtain a set of decoupled subsystems, on which we can easily analyze the convergence rate. By using Routh stability criterion from control theory, we derive the explicit formula of the optimal worst-case convergence rate and the corresponding parameters. We can see that the convergence rate of DIGing is slow even for the simplest objective functions, thus acceleration is necessary for general application. The proposed method can be viewed as the first step toward optimal parameter design of DIGing algorithm in solving general objective functions.

Keywords: Distributed optimization, Multi-agent systems
Abstract: This paper considers the distributed Nash equilibrium (NE) seeking problem for multi-cluster games in open networks, where players can freely join or leave the game with certain probabilities. In multi-cluster games, within a same cluster, players cooperate to optimize the cost function of the cluster, and different clusters compete, which can be seen as a noncooperative game. To track the time-varying NE of this formulated game timely, based on the inter- and intra- communication networks, a distributed projected gradient tracking algorithm under open networks is designed. Through rigorous theoretical analysis, we prove that the strategy profile sequence of players generated by the proposed algorithm linearly converges to a neighborhood of the time-varying NE in expectation. Finally, a numerical simulation of Cournot competition game is presented to illustrate the theoretical result.

Keywords: Distributed optimization, Multi-agent systems, Consensus
Abstract: In this paper, the predefined-time output-consensus distributed optimization is investigated for heterogeneous linear multi-agent systems. The proposed solution is an algorithm consisting of two main components: the first part uses an appropriate formulation of the output dynamics, while the second part represents a predefined-time controller which utilises sliding-mode control, zero-gradient-sum property, and predefined-time consensus functions to achieve consensus towards the global optimal solution to the optimization problem. This algorithm allows convergence under a user-defined upper-bound settling time. Finally simulation results are provided to prove its efficiency.

Keywords: Distributed optimization, Multi-agent systems, Cyber security networked control
Abstract: Distributed stochastic optimization enables multi-agent collaboration in applications such as distributed learning and sensor networks, but also raises critical privacy concerns due to the involvement of sensitive data. While existing privacy-preserving approaches often face limitations in balancing accuracy with efficiency, we propose a novel distributed stochastic gradient descent algorithm that integrates Paillier homomorphic encryption with heterogeneous and time-varying random stepsizes. The proposed algorithm provides inherent privacy protection against both internal honest-but-curious agents and external eavesdroppers, without relying on any trusted neighbors. Furthermore, we incorporate an attenuation factor to effectively mitigate quantization error induced by the encryption process, ensuring almost sure convergence to the optimal solution while maintaining privacy preservation. Numerical simulations demonstrate the effectiveness and efficiency of the proposed approach.

Keywords: Distributed optimization, Stochastic differential equations, Consensus
Abstract: This paper addresses distributed constrained optimization over networks with sporadic, asynchronous communication. We propose a Distributed Stochastic Primal-Dual Optimization Dynamics (Saddle-Point Flow) that integrates dual consensus dynamics driven by Poisson processes. Unlike standard approaches assuming synchronous updates, we model information exchange as Poisson-driven stochastic differential equations. For the case of a quadratic programming problem, we derive an explicit design rule linking a sufficient communication rate to the system's exponential convergence rate and ultimate accuracy. Our analysis proves that nominal performance can be recovered on distributed network topologies given sufficiently frequent communication, even in the presence of channel leakage.

Keywords: Cyber security networked control, Control over networks, Fault detection and diagnosis
Abstract: Networked control systems (NCSs) are vulnerable to faults and hidden malfunctions in communication channels that can degrade performance or even destabilize the closed loop. Classical metrics in robust control and fault detection typically treat impact and detectability separately, whereas the output-to-output gain (OOG) provides a unified measure of both. While existing results have been limited to linear systems, this paper extends the OOG framework to nonlinear NCSs with quadratically constrained nonlinearities, considering false-injection attacks that can also manipulate sensor measurements through nonlinear transformations. Specifically, we provide computationally efficient linear matrix inequality conditions and complementary frequency-domain tests that yield explicit upper bounds on the OOG of this class of nonlinear systems. Furthermore, we derive frequency-domain conditions for absolute stability of closed-loop systems, generalizing the Yakubovich quadratic criterion.

Keywords: Cyber security networked control, Data-driven control theory, Resilient networked control systems
Abstract: The concept of a security index quantifies the minimum number of components that must be compromised to carry out a stealth attack. This metric enables system operators to assess the security risk of each component and implement countermeasures accordingly. In this paper, we introduce a data-driven security index that can be computed solely from input/output data when the system model is unknown. We show a sufficient condition under which the data-driven security index coincides with the model-based security index, which implies that the exact risk level of each component can be identified solely from data. We also provide an algorithm for computing the data-driven security index.

Keywords: Cyber security networked control, Discrete event modeling and simulation, Control under communication constraints
Abstract: This paper addresses the security problem within an integrated attack-defense framework for discrete-time networked control systems with multiple sensors. Different from traditional indiscriminate denial-of-service (DoS) attacks, a novel data-importance-aware attack strategy is considered, which selectively targets critical sensor nodes to maximize impact. Unlike traditional duration-independent attacks, the attacks under consideration are modeled as a duration-dependent switching chain governed by a joint distribution function of current mode and its duration. A secure sliding mode controller is then synthesized to counter these kinds of attacks. Finally, the effectiveness of the proposed secure control strategy is validated through an RLC circuit example.

Keywords: Cyber security networked control, Fault detection and diagnosis, Control over networks
Abstract: This paper proposes an approach to opacity enforcement in parity space based remote monitoring systems. The true plant state is regarded as the secret to be protected. The attacker is not allowed to figure out the secret, even if he is able to eavesdrop both sensor output signals and control input signals transmitted over the network. The basic idea is to inject mask signals into the sensor output channels and the control input channels to alter signals transmitted over the network, so that the attacker can't obtain an accurate state estimate of the plant. In order to avoid any influence on the fault detection performance of the remote monitoring system, the mask signals are selected in the right null space of particular matrices related to the parameters of the remote monitoring system. An example is given to illustrate the proposed opacity enforcement approach.

Keywords: Cyber security networked control, Multi-agent systems, Consensus
Abstract: Privacy has become a critical concern in modern multi-robot systems, driven by both ethical considerations and operational constraints. As a result, growing attention has been directed toward privacy-preserving coordination in dynamical multi-robot systems. This work introduces a randomized scheduling mechanism for privacy-preserving robot rendezvous. The proposed approach achieves improved privacy even at lower communication rates, where privacy is quantified via pointwise maximal leakage. We show that lower transmission rates provide stronger privacy guarantees and prove that rendezvous is still achieved under the randomized scheduling mechanism. Numerical simulations are provided to demonstrate the effectiveness of the method.

Keywords: Cyber security networked control, Resilient networked control systems, Fault detection and diagnosis
Abstract: This paper investigates the vulnerability of discrete-time linear time-invariant systems to stealthy sensor attacks during the learning phase. In particular, we demonstrate that an adversary can inject attacks that mislead the operator into learning an {unstable} state-feedback controller. We further analyze attacks that degrade the performance of data-driven {H}_2 controllers, while ensuring that the operator can always compute a feasible controller. Numerical examples illustrate the effectiveness of the proposed attacks and underscore the importance of accounting for adversarial manipulations in data-driven controller design.

Keywords: Filtering and smoothing, Estimation and filtering
Abstract: This paper addresses the problem of online estimation of input time-delay in continuous–discrete (CD) linear systems. Motivated by practical applications where both continuous-time disturbances and measurement noise are present, we propose a robust delay and state estimator based on the Unscented H-infinity Filter (UHF). To extend the UHF framework to CD systems, we redesign the performance index to penalize the energy of continuous-time disturbances and derive the discrete-time state evolution under worst-case continuous-time disturbances, yielding a prediction step without probabilistic assumptions on delay variability or any disturbances. Through numerical simulations, we demonstrate that the proposed estimator achieves superior robustness and accuracy compared with the Unscented Kalman Filter (UKF).

Keywords: Kalman filtering, Estimation and filtering, Diagnosis of discrete event and hybrid systems
Abstract: In this paper, we provide a novel framework that enables a sensitivity-based observability test and state estimation algorithm for wind turbine power systems (WTPSs). The provided framework is the first of its kind in the literature, as it is able to deal with state- of-the-art WTPS models that are non-reduced, highly nonlinear differential-algebraic equation systems. Moreover, the framework includes nonsmoothness in both the dynamics and output functions to unify the operational conditions over different wind speed regions. We demonstrate the effectiveness of the proposed framework (thanks to the underlying tools from generalized derivatives theory) on a standard wind speed profile. We also illustrate how the proposed framework, by the utilization of robust observability analysis during nonsmooth transitions, enables accurate state estimation for cases when the conventional Extended Kalman Filter approach fails.

Keywords: Linear system identification, Distributed control and estimation, Control of networks
Abstract: Identifying parameters of fractional-order discrete-time dynamical networks from input-output data is crucial for modeling complex systems in neuroscience and biology. Despite existing methods, fundamental conditions ensuring structural identifiability based solely on network structure remain unexplored.

This paper establishes identifiability theory for fractional-order networks by proving that a network is structurally identifiable if and only if all its subsystems are identifiable. We introduce a graph-theoretical hierarchical algorithm that systematically partitions networks into identifiable subsystems via input-output reachability, enabling decomposition-based parameter learning with provable guarantees. Extensive simulation experiments validate that this approach preserves identifiability while achieving significant computational speedup, demonstrating scalability for large-scale systems.

Keywords: Modeling and estimation in agriculture, Sensing and perception in agriculture, Real time monitoring and control of environmental systems
Abstract: Aquaponics research is challenged by fragmented measurements, heterogeneous data formats, and missing consistent modeling frameworks. This paper proposes a digital infrastructure built on two pillars: (i) a unified protocol specifying parameters, frequencies, and formats, and (ii) a transient MATLAB model to simulate biomass growth and nutrient cycling. The framework demonstrates consistency with defined parameters, enables scenario analysis, and provides predictive capacity. The scenario analysis further shows that feeding intensity is the primary driver of growth, plant area limits nitrate removal, and nitrification efficiency governs residual load. These effects are synthesized into three compact key performance indicators (KPIs)—residual nitrate, uptake fraction, and feed–based efficiency—enabling comparative benchmarking and a control–ready basis for aquaponics systems. This is the first integrated data–model framework enabling reproducible calibration and providing a digital-twin foundation for aquaponics optimization.

Keywords: Physics informed and grey box model identification, Nonlinear system identification, Machine and deep learning for system identification
Abstract: This paper proposes a novel orthogonal-by-construction parametrization for augmenting physics-based input-output models with a learning component in an additive sense. The parametrization allows to jointly optimize the parameters of the physics-based model and the learning component. Unlike the commonly applied additive (parallel) augmentation structure, the proposed formulation eliminates overlap in representation of the system dynamics, thereby preserving the uniqueness of the estimated physical parameters, ultimately leading to enhanced model interpretability. By theoretical analysis, we show that, under mild conditions, the method is statistically consistent and guarantees recovery of the true physical parameters. With further analysis regarding the asymptotic covariance matrix of the identified parameters, we also prove that the proposed structure provides a clear separation between the physics-based and learning components of the augmentation structure. The effectiveness of the proposed approach is demonstrated through simulation studies, showing accurate reproduction of the data-generating dynamics without sacrificing consistent estimation of the physical parameters.

Keywords: Vehicle dynamic systems, Automotive system identification and modelling
Abstract: Integrating dynamics models into vehicle state estimation improves estimator performance. However, the inherent saturation characteristics of tires introduce significant state uncertainty under extreme conditions. This paper proposes an observability-driven adaptive vehicle velocity estimator. Based on a dynamic and combined tire model, an estimation framework is developed that fuses kinematic prediction and dynamics feedback. Then, an observability-driven adaptive mechanism is proposed, which adjusts the dynamic feedback weight according to the minimum singular value of the empirical Gramian matrix and supervises the estimation through innovation when kinematic prediction dominates. This mechanism enables the effective utilization of dynamic information while suppressing unreasonable feedback. Experiments on compacted snow under circular driving verify the effectiveness of the method, reducing PE and RMSE by 60.41% and 60.99%, respectively.

Keywords: Advanced process control, Process modeling, identification, and estimation techniques, Industrial applications of process control
Abstract: This paper builds on two recent papers which created a coherent formulation for generating proportional-integral-derivative (PID) controllers using a state-space formulation Abramovitch (2026a,b). Those papers showed the state structure, the estimator formulation, and the constraints on the estimator gains to allow us to generate a PID controller directly from a state-space structure. In this paper, we go further by enabling the use of a Kalman-Bucy filter Bryson and Ho (1975); Wikipedia (2025a); Bucy (1967) to generate the estimator gains, subject to the constraints shown in Abramovitch (2026a,b).

Keywords: Advanced process control, Industrial applications of process control
Abstract: Actuator saturation is a critical nonlinearity that degrades the performance of feedback control systems by inducing integrator windup and prolonging recovery from disturbances. Anti-windup strategies are designed to mitigate these effects, yet their performance depends strongly on controller design choices and tuning. This paper presents a novel tuning rule for the tracking time constant of the back-calculation anti-windup scheme for first-order-plus-dead-time processes and load disturbances, derived from an extensive optimization study that minimizes the Integral of Absolute Error (IAE) under varying levels of actuator saturation and disturbance conditions. Four practical tuning rules are presented to cover cases with different levels of information about the disturbance duration. Simulation results confirm that the proposed approach improves disturbance rejection while maintaining simplicity of implementation.

Keywords: Advanced process control
Abstract: In this paper we analyze the performance that can be obtained by adding an integer and a fractional double derivative action to a Proportional-Integral-Derivative (PID) controller for double integrator processes. In particular, the controllers are optimized in order to minimize the integrated absolute error (IAE) subject to constraints on the maximum sensitivity. Both the set-point following and the load disturbance rejection tasks are considered separately. A fragility analysis is also performed for the fractional double derivative order. Results show the trade-off between the increased complexity of the controller and the increment of the performance. Indeed, the use of a fractional double derivative action allows a significant performance improvement, especially for the load disturbance rejection task.

Keywords: Advanced process control, Industrial applications of process control, Control of multi-scale, distributed, and particulate systems
Abstract: A common requirement for automatic control of platforms in application areas such as renewable energy, electric vehicles, and robotics is the regulation of complex integrating systems in the form of position control. A standard control strategy is to decompose the complex integrating system's dynamics into a stable subsystem and a pure integrating system to form a cascade feedback control structure. This paper develops an approach to the automatic tuning of integrating systems using a cascade control system structure. The new design's main advantages compared to existing approaches include simplifying the auto-tuner for integrating systems and much improved closed-loop disturbance rejection performance for this control system class. Experimental results from application to an electro-mechanical system demonstrate the auto-tuner's efficacy and the superior closed-loop performance of the cascade PID control system.

Keywords: Advanced process control
Abstract: Feedforward control is an efficient strategy to compensate for measurable load disturbances. An ideal feedforward compensator is obtained as the ratio between the transfer function from the load disturbance to the process output and the transfer function from the control signal to the process output, with reversed sign. If applied, this ideal feedforward compensator will eliminate the response in the process output completely. The problem is that this compensator is not realizable in many cases, since the compensator may be non-causal, unstable, or improper. Tuning rules that take the first two cases, causality and stability, into account have been developed during the last decades, but the last problem of improperness has not been treated before. An improper transfer function can be made proper in two ways, by reducing the order of the numerator or by increasing the order of the denominator. This paper presents compensator structures and tuning rules that are based on both these approaches. The tuning rules are based on tradeoffs between performance in terms of IE and IAE values, and control signal activity in terms of magnitude of control signal peaks at step load disturbances. The paper includes analytical derivations as well as simulation examples.

Keywords: Advanced process control, Real-time optimization and control in chemical processes, Industrial applications of process control
Abstract: Advanced regulatory control (ARC), also known as advanced PID architectures, is emerging as a simple and robust way of controlling process with changing and possibly conflicting constraints, where it previously was believed - at least in academia - that model-based solutions, such as MPC, was the only effective solution. To illustrate this, ARC is applied on three case studies. The first is a gas-liquid separation process, where selectors and split-parallel control are combined to achieve bidirectional inventory control where the throughput manipulator moves automatically to the most optimal position. The two other case studies are on keeping acceptable air quality (CO2-level) and temperature in a room (in this case, a barn for cows). The CO2 and temperature constraints may be conflicting, leading to an hierarchical switching network of PID controllers.

Keywords: Adaptive control design, Applications of optimal control, Uncertain systems
Abstract: This paper presents a multi-loop control framework for a two-arm robotic manipulator with nonlinear dynamics and various sources of uncertainty. First, a neural network–based control policy is trained offline to drive the nominal dynamics to achieve near-optimal performance. To address discrepancies between the actual and nominal dynamics, which include modeling errors, disturbances, and actuator faults, a piecewise-constant adaptive law with radial basis functions is developed to estimate these uncertainties. An inner-loop control law then compensates for the estimated discrepancies, ensuring that the actual system behaves similarly to the nominal system under ideal conditions. By integrating the offline data-driven near-optimal policy with the online model-based adaptive control, the proposed framework guarantees near-optimal performance despite uncertainties and disturbances. Simulation results demonstrate that, under the multi-loop control scheme, the perturbed robotic manipulator maintains nearly the same tracking performance as in the nominal case.

Keywords: Adaptive control design, Sliding mode control, Robust control applications
Abstract: 4-DOF tower cranes are known to be challenging to control due to their complex dynamics. This paper proposes a control method for a 4-DOF tower crane that reduces the controller’s model dependency and demonstrates its effectiveness via experimental validation. Specifically, the proposed method utilizes only the mass matrix of the Euler-Lagrange (EL) dynamics and introduces an integral term into the auxiliary variable. We analytically prove that this design compensates for constant unmatched disturbances, such as crane calibration errors. The stability analysis is performed using the Lyapunov method and mathematical analysis.

Keywords: Adaptive control design, Uncertain systems, Application of nonlinear analysis and design
Abstract: This paper proposes a steering control mechanism for a soft robotic endoscope to autonomously navigate through the colon. A two-parameter adaptive steering controller, for which the visual servo problem of the soft robot is recast into the classical visual servo paradigm by introducing time-varying parameters, is proposed. The steering control guarantees an adjustable uniform ultimate bound for regulation error in the presence of disturbances and parameter variations. The controller is implemented with a navigation scheme to achieve autonomous navigation. Two experiments on set-point regulation and navigation inside a colon phantom, respectively, are presented. The results show that the proposed method can perform regulation accurately without spiral detours, and the navigation performance outperforms a proficient human operator in terms of accuracy and compliance.

Keywords: Control in system biology, Model predictive control, Applications of optimal control
Abstract: Interpatient variability is a main challenge in model-based closed-loop anesthesia, mainly due to poor modeling of peripheral compartments which are weakly informed by clinical data. This paper proposes a tube MPC framework that treats their influence as additive disturbances. Additionally, a new co-administration approach is developed, and a shrinking-horizon strategy is employed to reduce the computational burden of the MPC algorithm. The proposed controller is then evaluated against a baseline PID controller from the literature.

Keywords: Decentralized control, Control barrier functions and state space constraints, Large-scale and networked optimization problems
Abstract: In surveillance systems utilizing autonomous robots, both area exploration and tracking of detected moving targets are required. This paper presents a method which integrates persistent coverage control and cooperative tracking for multiple two-wheeled mobile robots. The proposed method enables a team of mobile robots to flexibly switch between exploration and tracking according to the situation. By employing the persistent coverage with cooperative tracking rule, it is shown that the proposed method maintains surveillance performance while responding to dynamic targets. Simulation and experimental results demonstrate that cooperative tracking improves coverage efficiency compared with single-robot tracking.

Keywords: Design methods for data-based control, Adaptive control design, Applications of optimal control
Abstract: Data-Enabled Predictive Control (DeePC) has recently emerged as a framework for controlling unknown systems from data. However, its performance relies on the relevance of the collected data, and as such, disturbances lead to inevitable errors. This paper addresses this problem by proposing an augmentation of DeePC using Model-Free Disturbance Observer with Online Modification (MFDOOM). The method corrects output predictions based on previous prediction errors using a dedicated continuously updated Hankel matrix. We compare our method, both theoretically and through simulation, to other recent algorithms designed for time-varying systems in the DeePC framework. It is shown that for disturbances that can be modeled as the output of an autonomous linear time-invariant system, this approach can reduce tracking error and online-update burden compared with existing online DeePC variants.

Keywords: Model predictive control, Design methods for data-based control
Abstract: Data-driven control techniques, such as DeePC, enable controller synthesis directly from data, but remain challenging for nonlinear or stochastic systems with limited data. Existing extensions, such as S-DeePC, address this through problem decomposition, but rely on indirect formulations. This paper presents F-DeePC, a fragmented DeePC approach that builds the constraint matrices directly and splits the prediction horizon into shorter fragments. The construction supports heterogeneous fragment lengths, while this paper focuses on the homogeneous case. The resulting formulation improves flexibility and data efficiency, yielding better tracking performance under limited data and nonlinear dynamics.

Keywords: Model predictive control, Learning methods for optimal control, Applications of optimal control
Abstract: Efficient energy management is critical for microgrids with renewable resources and energy storage systems. This paper proposes an imitation learning-based framework to approximate mixed-integer Economic Model Predictive Control (EMPC) for microgrid energy management. By training a neural network to imitate an expert EMPC policy offline, the framework enables real-time decision-making while bypassing the unpredictable computational costs of online mixed-integer optimization. To improve robustness, we apply noise injection to mitigate distribution shift. Furthermore, a constraint-tightening approach combined with a projection layer is introduced to guarantee recursive feasibility and constraint satisfaction. Simulation results show the learned policy achieves near-optimal performance while reducing computation time by one order of magnitude.

Keywords: Model predictive control, Learning methods for optimal control, Applications of optimal control
Abstract: To address the critical challenge of model mismatch in Model Predictive Control (MPC) for autonomous vehicle trajectory tracking, this paper proposes a novel hybrid offline-online Gaussian Process Regression MPC (HOO-GPR-MPC) framework. While GPR-enhanced MPC can improve model fidelity by learning unmodeled dynamics, existing approaches face a trade-off: offline-learned GPR models lack adaptability to new conditions, while online learning suffers from escalating computational costs that hinder real-time application. Our method synergistically integrates offline pre-training with efficient online updates to overcome these limitations. The core contributions are threefold: first, an active data management strategy based on predictive variance to preserve a compact and informative online dataset; second, a performance-based dynamic weight assignment mechanism to adaptively fuse offline and online GPR models; and third, a real-time implementable and theoretically supported HOOGPR-MPC framework that integrates dynamic sparsification and asynchronous updates with formal guarantees on bounded uncertainty, bounded weight variation, recursive feasibility in probability, and ISS in probability.

Keywords: Model predictive control, Nonlinearity learning from data
Abstract: We consider the control of dynamical systems subject to constraints of arbitrary shape. In particular, we are interested in cases where some of the constraints are not described by a smooth function but are instead specified by a mix of functions and logical statements, a lookup table, or an oracle that returns whether a queried point is safe. We propose to use a Support Vector Machine (SVM) classifier to approximate the constraint boundaries and use it to define a smooth nonconvex optimal control problem (OCP). We derive tightening bounds on the classifier to ensure safety and investigate a class of kernels that lead to the OCP being a Difference-of-Convex (DC) programming problem. Moreover, we show a natural difference of convex function decomposition for the Gaussian Radial Basis Function. The approach is numerically validated for a planar robotic manipulator with obstacle avoidance constraints.

Keywords: Model validation, Digital implementation, Parametric optimization
Abstract: Experimental digital twins (EDTs) for real-time industrial control must handle sparse data, stable long-horizon prediction, and nonstationary dynamics. Static or purely data-driven models rarely satisfy all three at once. We propose an evolvable physics-informed neural network (PINN) framework built around three modules: Dymo-PINN injects physical laws into system identification; Tempo-CL, a contrastive fine-tuning scheme, curbs error accumulation in iterative prediction; and ResAdapter, a lightweight adapter that enables online adaptation without full retraining. On permanent magnet synchronous motor (PMSM) benchmarks, the framework raises R^2 by up to 1.82times under severe data scarcity and holds MSEs to 10^{-1}--10^{-2} under parameter drift, surpassing conventional PINN, contrastive, and incremental learning baselines in data efficiency, robustness, and cross-configuration generalization.

Keywords: Observer design, Nonlinearity learning from data, Model validation
Abstract: This work develops a control-oriented hybrid modeling framework for concrete three-dimensional (3D) printing that couples first-principles pump dynamics with machine-learning (ML) estimators of fresh slurry density (rho) and viscosity (mu) to dynamically predict the slurry mass flow rate (dot{M}). A physics-guided feature set is screened using correlation, mutual information, and permutation importance and trained under grouped cross-validation across mixes (36 nominal mixes; 97 density and 195 rheology points). In the density predictor pipeline, a non-negative least-squares (NNLS)-stacked ensemble using Ridge, Extra Trees (ExT), and Random Forest (RF) outperforms single learners (coefficient of determination (R^2) approx 0.73). For viscosity, high-capacity learners dominate; the best performance is achieved by a ridge-meta stack trained over the full base set (R^2 approx 0.83). Unlike prior screw-extrusion flow-control work, validated on a single rig without a systematic sweep of mix compositions and reporting mean absolute error (MAE) of approx 6.7%, this study targets progressive-cavity pumping and learns rho and mu from a multi-mix dataset for dynamic, control-oriented flow prediction. Embedding the learned rho and mu maps within the pump model yields simulated flow-rate curves that closely follow laboratory measurements across water-accelerator settings and motor speeds, with MAE on the order of 1--2% of mean flow. The predictor is interpretable and low-latency (milliseconds per evaluation) and integrates with Simulink for feedforward planning, soft sensing, and closed-loop use. The workflow provides a reproducible template, from data preparation and feature screening to model training and deployment, for hybrid modeling in large-format additive construction.

Keywords: Linear parameter-varying systems, Robustness analysis, Control of uncertain LPV systems
Abstract: This paper investigates the lower bound of the ℓ2 induced norm of discretetime linear systems subject to rate-bounded time-varying parameters. This paper proposes a numerical method combining the lifting technique for discrete-time systems and a numerical optimization to find a worst-case parameter sequence that attains the lower bound for the ℓ2 induced norm. This approach successfully identifies a periodic worst-case parameter sequence and the corresponding input signals, attaining lower bounds that closely match LMI-based upper bounds. A numerical example suggests that the worst-case parameter variation pattern and input signals are nontrivial to identify.

Keywords: Linear parameter-varying systems, Lyapunov methods, Passivity-based control
Abstract: We show that a common Lyapunov matrix exists for the convex combination of two Hurwitz matrices if and only if the intersection of the set of strict Lyapunov matrices for one matrix and the set of non-strict Lyapunov matrices for the other is nonempty. This simple relaxation is useful for the convergence analysis of the augmented primal-dual gradient flow for constrained optimization problems with affine inequality constraints, which can be viewed as a polytopic linear parameter-varying (LPV) system driven by the active-constraint selector. Under a relaxed strong convexity condition, exponential convergence is proved for the LPV system. The analysis can further be extended to the integral quadratic constraints (IQCs) framework for LPV systems to facilitate numerical search of the convergence rate.

Keywords: Linear parameter-varying systems, Control of uncertain LPV systems, Robust control applications
Abstract: This paper proposes a new Scheduling-Informed Linear Parameter Varying (SI-LPV) approach extended with scheduling-informed performance criteria. This method is able to handle actuator nonlinearities and unknown dynamics in linear systems, and as a novel element, the precise physically-based formulation of the performance criteria can be avoided. The SI-LPV approach uses Kolmogorov-Arnold representation theorem with ultra-local model formulation to achieve the control-oriented SI-LPV form. The paper presents how the measured performance signals can be built in the control. The effectiveness of the resulted controller is illustrated through examples, focusing on vehicle control problems, illustrating the effectiveness through longitudinal and vertical control design.

Keywords: Linear parameter-varying systems
Abstract: The trajectory tracking problem for autonomous vehicles can be enhanced by means of an effective unification of longitudinal and lateral control. To this end, Nonlinear Model Predictive Control (NMPC) is highly suitable as it inherently handles multi-objetive problems and accounts for process constraints. However, a key challenge in implementing NMPC schemes lies in the involved computational burden, which can impede systems subject to stringent real-time constraints. This issue directly affects the optimization feasibility, its overall effectiveness, and closed-loop stability. To overcome these limitations, we employ a Linear Parameter-Varying (LPV) representation of the vehicle's dynamic model. This article proposes an integrated LPV MPC solution for autonomous vehicle trajectory tracking. Accordingly, we propose an integrated LPV MPC establishing guarantees for both recursive feasibility and closed-loop stability via the formulation of parameter-dependent Linear Matrix Inequalities. The proposed approach is validated through comparative simulations using a high-fidelity nonlinear dynamic model of a scaled vehicle, demonstrating similar tracking performances compared to NMPC but strongly reducing computation time, making it suitable for real time applications.

Keywords: Linear parameter-varying systems, Learning methods for optimal control, Robust control applications
Abstract: The paper presents a novel combination method for integrating an LPV-based controller with reinforcement learning. The main idea behind the combination is that the agent learns the scheduling parameters of the LPV observer to cope with the unmodeled or uncertain dynamics of the considered system. The RL aims to simultaneously minimize the error of the observer and the tracking performance of the controller. In this way, the exploration of the nonlinear system and the stability of the closed-loop system can be guaranteed. The proposed method is validated through two vehicle-oriented control problems, namely the trajectory tracking of ground vehicles and traction control of trains.

Keywords: Linear systems, Nonlinearity learning from data, Design methods for data-based control
Abstract: This paper presents a control architecture in which a Linear Quadratic Regulator (LQR) is integrated with a machine learning-based nonlinear compensator realized through a Kolmogorov-Arnold Network (KAN). The goal of the proposed approach is to achieve a balance between guaranteed closed-loop stability, computational efficiency, and enhanced control performance. First, a Reinforcement Learning (RL) framework is used to learn an effective nonlinear control policy across the entire operating range of the system. Then, the trained RL-based control method serves as the teacher network for the KAN-based algorithm. During the knowledge distillation, gradient regularization and coefficient constraints are applied to achieve a smooth and Lipschitz-bounded neural network-based controller. Stability is guaranteed for the combined LQR-KAN closed-loop system through the computation of the maximum allowable Lipschitz constant of the neural network. The proposed control structure is validated on a double inverted pendulum system. The results show that the combined KAN-based controller achieves nearly the same performance level as the RL-based method, while the Lipschitz constant and the complexity of the network are significantly reduced. This property directly supports simple stability analysis and reliable real-time implementation.

Keywords: Sliding mode control, Linear systems, Linear parameter-varying systems
Abstract: This paper presents a methodology for high-order sliding-mode (HOSM) control of perturbed single-input linear time-varying (LTV) systems. The considered class satisfies a differential uniform controllability (DUC) property characterized through an extended controllability matrix. A key contribution is the construction of a state immersion that guarantees the existence of a sliding output with full relative degree. The methodology enables the implementation of HOSM controllers acting on the extended state, thereby ensuring finite-time stability and, under suitable parameter selection, fixed-time stability.

Keywords: Sliding mode control, Lyapunov methods
Abstract: This paper investigates the possibilities of application of generalized homogeneity for the finite-time stabilization in linear systems. The generalized dilation is derived that establishes homogeneity of a purely oscillating linear system and a serial connection of a linear oscillator and an integrator. We present a theoretical framework based on Implicit Lyapunov Functions (ILFs) to analyze finite-time stability of linear systems. Building on these results, a novel nonlinear control law is developed to achieve robust stabilization of linear systems.

Keywords: Sliding mode control, Lyapunov methods
Abstract: Continuous homogeneous state-feedback control of a disturbed chain of integrators is considered for the exemplary second-order case. The control takes full state information corrupted with measurement noise. Within a high-gain setup, we choose the controller gains that are optimal w.r.t. the estimated effect of the noise and a matched disturbance on the control error. This is quantified in terms of the homogeneous Lp-gain, a generalization of the classical Lp-gain for homogeneous systems recently introduced. To estimate the homogeneous Lp-gain, we construct a solution to the associated Hamilton-Jacobi-Inequality based on a Lyapunov function. This estimate is minimized leading to an optimal gain-selection that balances the effect of noise and disturbance on the control error of interest. The estimation- and minimization procedures are evaluated numerically and applied to a laboratory experiment.

Keywords: Sliding mode control, Stability of nonlinear systems, Lyapunov methods
Abstract: The paper deals with the problem of robust stabilization of a nonlinear plant. The method of homogeneous invariant/attractive ellipsoids is developed for a robust homogeneous stabilizer design. Similarly to linear case, the control tuning is based on solving of a semi-definite programming problem. Such a design is shown to be possible for a class of nonlinear systems satisfying a special homogeneous conic constraint, which means that the nonlinear system, in some sense, is close to a linear controllable system. The optimal tuning for a class of affine-in-control systems and homogeneous sliding mode control systems is studied. Theoretical results are supported by numerical simulations.

Keywords: Sliding mode control, Stability of nonlinear systems, Lyapunov methods
Abstract: A fast convergence in a fixed-time of solutions of nonlinear dynamical systems, for which special requirements are satisfied on the derivative of a quadratic function calculated along the solutions of the system, is proposed. The conditions for the system solutions to converge to zero and to a given region within a fixed-time are obtained. To achieve fast convergence, a negative power is applied to the derivative of a quadratic function within a specific time interval during the evolution of the system. The application of the proposed results to the design of control laws for arbitrary order linear plants using the backstepping method is considered. All the main results are accompanied by numerical modelling and a comparison of the proposed solutions with some existing ones.

Keywords: Optimal control theory, Applications of optimal control, Sliding mode control
Abstract: This paper presents a finite-time stabilization method for the integrator chain via optimal control techniques by introducing a tailored cost function. We analyze key properties of the resulting value function to derive the corresponding Hamilton-Jacobi-Bellman equation. Numerical simulations illustrate the effectiveness of the approach.

Keywords: Linear systems, Linear time-delay systems, System structure and control
Abstract: Low-order stable controllers are designed for plants satisfying the parity interlacing property with constrained right half plane poles and zero structures, for three different classes. A lower bound of the delay margin achievable by the proposed controllers is also obtained. There are several free parameters in the low-order strongly stabilizing controller design procedure. It is shown, via examples from the literature, how these parameters affect the delay margin and the sensitivity of the feedback system.

Keywords: Linear time-delay systems, Impulsive linear systems, System structure and control
Abstract: This work provides a constructive sufficient solvability condition for the disturbance decoupling problem with stability in a special class of impulsive systems with impulse delays. The so-called reset-delayed linear systems are impulsive systems in which, at the jump times, some or all state variables are brought back to the values they had a given amount of time, the reset delay, before the jump. The disturbance decoupling problem addressed consists of finding a state feedback such that the output of the compensated reset-delayed linear system is unaffected by the disturbance and the closed-loop dynamics is globally asymptotically stable provided that the time interval between any two consecutive jump times is sufficiently large — dwell-time global asymptotic stability. Under the assumption that the flow dynamics is stabilizable, the condition is expressed in geometric terms using a novel subspace that has appropriate structural properties with respect to the flow and jump dynamics and is internally stabilizable subject to the dwell-time constraint.

Keywords: Linear time-delay systems, Linear functional systems, Lyapunov methods
Abstract: This paper extends recent stability criteria in terms of delay Lyapunov matrices to the case of linear systems with both pointwise and distributed delays. The approach is based on Lyapunov–Krasovskii functionals with prescribed derivatives, where the kernels are approximated by piecewise linear matrix functions. Then, Gu's discretized Lyapunov functional method is employed to transform the approximated functional to a quadratic form structure, whereas finite dimension of the stability test is calculated by quantifying the approximation error. The resulting stability criterion maintains a convenient structure that depends on evenly spaced pointwise values of the delay Lyapunov matrix. This formulation leads to a finite computation process, reduces conservatism, and allows an efficient numerical implementation.

Keywords: Linear time-delay systems, PDEs for time delay systems, Lyapunov methods
Abstract: In this paper, the problem of constructing the delay Lyapunov matrix for linear periodic time delay systems with a constant delay is considered. It is known that, in the case when the delay is an integer multiple of the period of the system matrices, the delay Lyapunov matrix is given by a solution of a matrix hyperbolic PDE problem defined on a strip. Our approach is based on constructing the matrix fundamental solution for this hyperbolic PDE problem. The representation formula for the delay Lyapunov matrix is then derived. It depends on the fundamental solution and unknown initial matrices of the PDE problem. Finally, a non-standard boundary condition given in the PDE problem is transformed to a vector Fredholm integral equation of the second kind in order to determine initial matrices.

Keywords: Positive linear systems, Linear functional systems, Uncertain systems
Abstract: This paper deals with the stability analysis of linear continuous-time positive difference equations with multiple time-varying delays. We consider the general case of time- varying matrices, and introduce novel sufficient conditions for global exponential stability with guaranteed exponential decay rate.

Keywords: Nonlinear time-delay systems, Sampled-data/digital control, Stability of nonlinear systems
Abstract: In this paper, the digital implementation of continuous-time stabilizers is addressed for nonlinear time-delay systems. In particular, for nonlinear systems with state delays and affected by known exogenous disturbances, it is shown the existence of a suitable fast sampling and of an accurate quantization of both input and output channels such that the digital event-triggered implementation of continuous-time global asymptotic stabilizers ensures the semi-global practical stability property of the corresponding closed-loop system with arbitrarily small final target ball of the origin. In the theory here developed, aperiodic sampling and the non-uniform quantization of both input/output channels are allowed. An application is presented to validate the results.

Keywords: Industry X.0 for production and logistics, Smart production and logistics in manufacturing, Manufacturing engineering and management
Abstract: Drawing on empirical insights and real-life examples from the Autonomous Factory project within the NXTGEN program, a Dutch national initiative in the high-tech sector, this study analyzes the barriers hindering Smart Industry (SI) adoption. It examines how these barriers can be alleviated through collaboration, shared expertise, and financial support. The paper then proposes a conceptual framework that links sector characteristics, barriers, and interventions that are often examined in isolation, and shows how coordinated actions can overcome the interconnected barriers of SI adoption in high-tech manufacturing. It also offers practical insights for policymakers and industrial managers seeking to accelerate SI transformation.

Keywords: Intelligent manufacturing systems, Systems-of-systems, Manufacturing engineering and management
Abstract: Concurrent Engineering (CE) is increasingly adopted to address mass customization in manufacturing. Central to CE are the multi-disciplinary collaboration and cross-domain integration, which make Model-Based Systems Engineering (MBSE) a promising approach in this regard. Although MBSE is well established in industries such as aerospace, it is not yet tailored to the specific needs of discrete manufacturing. This paper reviews the existing literature on MBSE for CE to advance the understanding of its applicability in manufacturing systems, with particular focus on formal modeling and model-based analysis. The findings support the development of a reference framework that identifies key elements and pillars required to enable CE through MBSE. Building on this framework, a research agenda is proposed, highlighting the need for integrated formal models that encompass product, process, and production system domains, including their features and interdependencies. Advancing in this direction will enhance the reusability, scalability, and flexibility of manufacturing systems, supporting more effective CE.

Keywords: Industrial artificial intelligence, Manufacturing plant simulation, control and optimization, Intelligent manufacturing systems
Abstract: The kilning phase in malt processing is essential for halting germination and inducing flavor, color, and aroma development in barley. Optimizing the kilning phase presents challenges due to its dynamic nature and high energy demands. Traditional control methods often result in sub-optimal outcomes. We propose a novel framework based on Reinforcement Learning (RL) to develop customized recipes for malt kilning, aiming to improve outcomes in energy consumption, duration, cost, and CO2-eq emissions. Our approach leverages RL algorithms to generate optimized recipes based on initial conditions and specific key performance indicators (KPIs). Initial results show improvements over existing methods in cost and time, indicating the potential for broader implementation and optimization in the malt processing industry.

Keywords: Cyber-physical production systems, Industrial artificial intelligence, Intelligent manufacturing systems
Abstract: Abstract: Digital Twin applications in manufacturing increasingly require autonomy in decision-making, yet existing approaches fail to develop proper prescriptive capabilities that could enhance production planning and control activities. To cope with such limitations, this paper proposes a dual-agent framework that embeds a Reinforcement Learning agent directly within the Digital Twin, while maintaining an external Genetic Algorithm for long-term planning. The internal agent operates with continuous access to production resources states during simulation execution, while the external agent optimizes schedules across extended time horizons. The framework provides guidance for developing prescriptive Digital Twins capable of autonomous scheduling decisions in dynamic manufacturing environments. Validation within a highly automated assembly line demonstrates that the dual-agent approach outperforms traditional approaches in average lead time reduction, achieving deviation from the optimal below 2% and a 33% improvement over GA alone.

Keywords: Industrial artificial intelligence, Intelligent manufacturing systems, Smart production and logistics in manufacturing
Abstract: The interpretation of complex industrial documentation remains a challenge for Artificial Intelligence in manufacturing. This paper presents an evaluation of state-of-the-art Vision Language Models (VLMs) (GPT-5, Gemini 2.5 Pro, Claude Sonnet 4.5, Claude Opus 4, Qwen VL Max, Llama 4 Maverick) as support assistants. Using a dataset from an industrial machine programming manual, performance on images, diagrams, graphs and tables at different cognitive difficulty levels is tested. A multidimensional metrics methodology is employed. Results show strong performance on images and schematics but notable degradation on quantitative elements, highlighting the need for dedicated benchmarks for VLMs in industrial contexts.

Keywords: Data-driven and AI-based modelling of production and logistics, Intelligent manufacturing systems, Industrial artificial intelligence
Abstract: Final quality testing in automotive manufacturing is crucial to ensure standards before delivery. This study applies sequential pattern mining to sequences of defects and reworks, extracting frequent patterns as features in classification models that predict final test outcomes. The approach enables a reduction in input data dimensionality and the selection of representative variables. Results show that using mined patterns as predictors lowers misclassifications of defective vehicles compared to using all variables, with only a slight decrease in overall accuracy. This method enhances the efficiency and reliability of final quality assurance processes.

Keywords: Cyber-physical-social systems in enterprises, AI-based enterprise systems, Enterprise interoperability
Abstract: Industrial environments increasingly rely on collaboration between humans and AI-enabled agents. Effective teamwork requires aligning how agents perceive situations, plan actions to pursue goals, and adapt to changing conditions, yet existing systems lack mechanisms for cross-agent cognitive processes coupling. This paper presents a conceptual cognitive agent model that formalises cognitive coupling through eight components: Input, Process, Output, State, Value, Memory, World Model, and Goal. The model abstracts how agents coordinate and co-regulate their cognitive cycles, providing a basis for analysing distributed cognition and designing cognitively interoperable human–AI systems.

Keywords: Robotics in manufacturing systems, Cyber-physical-social systems in enterprises, Human-technology integration in manufacturing
Abstract: Human-Robot Collaboration (HRC) requires robots to interpret task semantics, understand human actions, and adapt autonomously when humans deviate from expected workflows. This paper describes how the cognition and simulation functions of a Cognitive Digital Twin (CDT) can be implemented by combining ontology-based reasoning with the CLARION cognitive architecture. A modular ontology (AI4C2PS) encodes tasks, affordances, capabilities, and abilities, enabling SWRL-based inference to determine when a robot can perform an action. These inferences are fed into CLARION, where the explicit subsystem handles semantic knowledge and the implicit subsystem learns procedural skills through Q-learning in simulation. A collaborative screwing scenario demonstrates how a cobot that was not initially programmed for screwing can infer the helical-motion capability and learn to execute it autonomously.

Keywords: Large-scale complex systems, Complex dynamic systems, Interconnected dynamical systems
Abstract: As urban heat island (UHI) effects—where urban areas experience significantly higher temperatures than surrounding rural regions—intensify under climate change, urban heat mitigation is becoming a central challenge for cities. Urban microclimate‑sensitive planning, however, still relies on fragmented tools, specialized expertise, and workflows that are difficult to integrate into early design phases. Generative AI offers new opportunities to ease this complexity, but its practical use in urban microclimate improvement planning remains largely unexplored. This paper presents GAIA, a conceptual generative‑AI planning assistant built on a multilayer architecture that integrates LLM‑based reasoning, geospatial context, domain‑knowledge retrieval, microclimate simulation feedback, and digital‑twin visualization. We detail the architectural design and illustrate its operation through a planner‑centred workflow. The concept provides a technically grounded foundation for next‑generation, AI‑supported urban microclimate planning.

Keywords: AI-based enterprise systems, Cyber-physical-social systems in enterprises, Enterprise interoperability
Abstract: As enterprise systems evolve toward more autonomous, adaptive, and context-aware capabilities, understanding how humans and systems can effectively interact, co-operate and collaborate is critical. This paper proposes an interactive cognition framework for constructing human-like cognition models, following the human-like cognitive processes, but also reasoning mechanisms, as well as methods for interpreting human intent, adapting machine behavior, creating common understanding or mindset, and fostering trust and transparency. By integrating large language model (LLM) with mathematical computation models of physical entities and enterprise business operation processes, and traditional optimization and machine learning algorithms, autonomous agents are developed for human-systems collaborative decision-making. A real-world industrial application case study is presented to illustrate the method for developing the interactive cognition agents, and specify the challenging factors and problems that should be considered and solved in the future.

Keywords: Sustainable and circular supply chain and production, Supply chain and logistics engineering, simulation and optimization, Sustainable and circular manufacturing systems
Abstract: Carbon Capture and Utilization (CCU) is a key pathway to decarbonize the chemical industry, but its economic viability strongly depends on policy and market conditions. This paper presents the optimal design of a CCU supply chain converting biogenic CO2-to-alcohols-to-olefins, aiming to minimize total cost of the supply chain, and integrating environmental performance. The optimal CCU system is evaluated by applying market-based mechanisms to quantify the impact of carbon pricing, hydrogen subsidies, product credits as demand pull instruments. Product Price Parity Index and Market Viability Index metrics are used to benchmark competitiveness against conventional fossil-based olefin production route. Results show that green olefins remain economically uncompetitive under current UK market conditions. However, a mix of mechanisms improve economic performance, enabling competitiveness with fossil-based route. This framework supports quantitative, policy-aware decision-making for CCU deployment.

Keywords: Advanced process control, Industrial applications of chemical process control
Abstract: Chemical processes are overseen by operators through human-machine interfaces (HMIs). Advanced control schemes such as model predictive control (MPC) are increasingly used to manage these complex chemical processes. However, MPC is perceived as a complex technique. This makes it non-intuitive for the operators, prompting them to manually override the controller. This work aims to understand human interaction with these advanced systems, especially in the presence of abnormal situations. We use a reactor-separator system, equipped with HMI to mimic an industrial control system and an eye tracker to analyze human interactions with the control system. Human factors study is performed with student participants who play the role of plant operators in our study. A comparison is made with the conventional system that involves only PID control. Results show their propensity to manually turn off MPC when faced with abnormal situations. However, compared to conventional control, participants in our study show a proactive approach when dealing with MPC, an improved performance and a clear effect of learning with multiple repetitions. Finally, analysis of dwell duration provides new insights into their perception of advanced control systems.

Keywords: Advanced process control, Industrial applications of chemical process control, Industrial applications of process control
Abstract: This paper investigates the tracking control problem for fully actuated nonlinear systems with a constant input delay. We combine the fully actuated system (FAS) approach with a pseudo predictor strategy. This strategy predicts future tracking errors by integrating a user-defined linear stable error dynamics, thus avoiding the instability arising from the prediction based on the potentially unstable open-loop system. We then establish a two-layer stability analysis framework from an input-to-state stability (ISS) perspective. The outer-layer tracking error system is input-to-state stable with respect to the inner-layer prediction bias. Based on this property, the prediction bias dynamics can be separated from the tracking error dynamics, and the asymptotic stability of the tracking error follows from the prediction bias being stabilized by Lyapunov-Krasovskii (LK) theory. Furthermore, the effectiveness of the proposed method is verified through its application to a two-stage chemical reactor.

Keywords: Advanced process control, Industrial applications of process control, Interaction between design and control in processes
Abstract: This paper proposes a novel dynamic event-triggered control (ETC) scheme for the continuous stirred tank reactor (CSTR) system with prescribed performance based on the fully actuated system approach (FASA) framework. First, a new prescribed-time performance function (PTPF) is designed to encode transient bounds and convergence time. Subsequently, the high-order error FAS model under the PTPF is derived, which enables explicit cancellation of nonlinear terms and closed-loop eigenstructure assignment. Based on this, a FASA-based dynamic ETC strategy with prescribed performance is established, in which an enhanced error-dependent triggering mechanism is employed to improve communication efficiency without sacrificing control performance. Rigorous analysis shows that the tracking error is confined to a prescribed set within the prescribed time and that Zeno behavior is excluded.

Keywords: Advanced process control, Real-time optimization and control in chemical processes, Machine learning and artificial intelligence in chemical process control
Abstract: This paper proposes a real-time optimisation (RTO) framework that uses AI-based surrogate models integrated with the advanced regulatory control (ARC). The main contribution is a method that employs a vertically decomposed control architecture, in which the RTO is formulated using steady-state surrogate models for both objectives and constraints. The RTO layer is responsible for guiding advanced regulatory control (ARC) to the optimal point subject to constraints, and the ARC layer coordinates the PID controllers during dynamic behaviour. The case study is an Electric Submersible Pump (ESP) system controlled with an ARC scheme. Results show effective constraint handling, operation within feasible regions, and adaptability to shifts between production and efficiency maximisation—antagonistic goals in ESP operation. The method is computationally efficient, resulting in a 99% reduction in processing time compared to differential equation methods. The RTO also re-optimised operating conditions following unmeasured disturbances. From a practical standpoint, the results indicate that real-time implementation can be more efficient and requires less computational effort.

Keywords: Industrial applications of chemical process control, Process modeling, identification, and estimation techniques, Industrial applications of process control
Abstract: To continue to support technological advancements in integrated circuits, semiconductor vertical furnace, which is widely used for oxidation, layer deposition, and annealing process, needs fast and high-precision temperature control even more. Conventional model-based control methods use a predetermined reference trajectory, which limits its speed during a temperature rise/fall process. Trajectory generation methods based on linear-time-invariant (LTI) models find difficulty due to the temperature dependency of the furnace. Therefore, the aim of this paper is to propose a control framework to generate fast trajectory which can cope with temperature dependency. This can be achieved by using heat capacity as a model which can be calculated empirically. Experiments on a full-size semiconductor vertical furnace verified its performance, being able to track the average temperature from 300°C to 400°C with nearly 50% time reduction compared to the control with a predetermined reference trajectory.

Keywords: Machine learning and artificial intelligence in chemical process control, Batch and semi-batch process control, Advanced process control
Abstract: This work introduces symbolic regression for controlling crystallization, presenting a robust methodology for controller development. Symbolic regression generated an equation to compute optimal control actions from temperature, supersaturation, and error measurements. The approach was tested in paracetamol batch crystallization in ethanol to regulate mass yield and crystal size by manipulating temperature. Its performance was compared to a nonlinear model predictive controller (NMPC) using a population balance model (PBM) as the internal model and a feedforward neural network trained with the same dataset. All approaches achieved successful control, but machine learning methods required a lower computational cost. Considering disturbance, plant model mismatches and noise, symbolic regression maintained variables near set-points with fewer temperature changes.

Keywords: Distributed optimization for smart grids, Control and management of energy systems, Energy management systems
Abstract: As distributed energy resources (DERs) penetrate distribution networks increasingly, the system becomes more unbalanced and thus, voltage issues arise frequently. To maintain voltages within a valid range while tolerating slight violations for short periods, voltage-optimization controls have been implemented to address voltage issues. Rather than linearizing the AC power flow equations to control the voltage, a deep deterministic policy gradient (DDPG) reinforcement learning (RL) algorithm is proposed to output regulator setpoint controls continuously. Such an approach offers real-time control once the models are trained for precise regulation without linearization errors. An efficient three-phase unbalanced power flow solver ensures high-fidelity RL environment during training. The algorithm is validated on the IEEE 13 and 123 bus systems for a one-hour snapshot control and demonstrates the effectiveness of the RL learned policy in minimizing voltage violations under varying load condition, promising extended controls, such as capacitors in sequential operation.

Keywords: Forecasting of power supply and demand, Solar energy, Energy management systems
Abstract: Accurate prediction of photovoltaic power generation is essential for maintaining grid stability and optimizing energy management. However, high nonlinearity, uncertainty and complex timing dependence of its power output make accurate prediction challenging. To this end, this paper proposes a dynamic modular reconstruction deep echo state network (DMRDESN). Historical data is first mapped to a high-dimensional state space through the reservoir. Then, according to the dynamic state of the reservoir neurons, similar neurons are grouped by the layer clustering method to form a modular sub-reservoir. The central neuron is determined in each module and interacts with other sub-reservoirs. Then, according to the modular structure, the weight of the reservoir is systematically reconstructed to enhance its representation ability. Experimental results show that the proposed DMRDESN model is tested on Mackey-Glass system and photovoltaic power generation data sets, and its prediction accuracy and stability are better than other prediction models.

Keywords: Power systems stability
Abstract: 大规模可再生能源资源的整合，导致系统惯性和阻尼减少，且 导致严重低频风险增加 振荡。网格成形（GFM）转换器的出现包括 这是一种增强系统惯性的有前景解决方案。然而，GFM转换器也可能参与，并且有可能 放大电力系统固有的振荡模式。要处理 本期论文提出基于库普曼算子的方案 用于GFM转换器的阻尼控制器。非线性 功率系统的动力学用线性表示 通过动态模式分解（DMD）识别模型，库普曼算符的数值实现。基于 该模型被识别为线性二次调节器（LQR） 用于设计通过GFM增强系统阻尼 转换器。在区域间进行的模拟研究 电力系统测试基准测试验证 确定了模型，并展示了该模型的有效性 拟议的阻尼控制器。

Keywords: Power systems stability, Control and management of energy systems, Control and optimization for sustainability and energy systems
Abstract: This paper proposes an optimization-based power flow calculation that guarantees small-signal stability for inverter-integrated power systems comprising synchronous generators, grid-forming (GFM), and grid-following (GFL) inverters. We demonstrate that the equilibrium of such a system can be determined using an equivalent system composed solely of GFM and GFL inverters. Based on this equivalence, we formulate the power flow calculation as an optimization problem by leveraging the convexity of the energy function. The solution is constrained within the convex domain of the energy function, thereby guaranteeing small-signal stability. Numerical simulations validate the effectiveness of the proposed method.

Keywords: Power systems stability, Energy market, Electrical distribution systems
Abstract: This paper proposes an energy function-based small-signal stability constrained optimal power flow (EF-SSSC-OPF) calculation for a homogeneous lossy power system with a radial structure. The energy function is defined for a lossless power system, which is equivalent to a homogeneous lossy power system. The EF-SSSC is formulated as an inequality constraint requiring the minimum eigenvalue of the energy function’s Hessian to exceed a user-specified value. An alternative SOCP relaxation linearizes the EF-SSSC and convexifies the OPF, thereby formulating the EF-SSSC-OPF as a convex optimization problem. Numerical simulations demonstrate the effectiveness of the proposed method in ensuring small-signal stability.

Keywords: Power systems stability, Power electronics
Abstract: The development of decentralized stability conditions has gained considerable attention due to the need to analyze multi-agent network systems, such as heterogeneous multi-converter power systems. This paper proposes a geometric decentralized stability condition based on Davis-Wielandt (DW) shell and its projections, which provides a geometric interpretation of the small-gain and small-phase theorems and enables decentralized stability analysis of power systems. It serves as a visualization method to understand the closed-loop interactions and assess the stability of large-scale network systems in a scalable and modular manner.

Keywords: Control and management of energy systems
Abstract: Transmission system operators in the natural gas sector, confronted with increasing challenges arising from the ongoing energy transition, require advanced optimization and simulation tools to support informed decision-making and efficient management of network infrastructure. This paper presents a methodology for developing a digital twin of a real gas transmission network. The proposed framework enables state estimation and data reconciliation to achieve a coherent and systematic representation of the physical system’s behavior. Preliminary validation has been conducted using synthetic data, with future work focusing on the application and validation of the approach using real operational data.

Keywords: Control and management of energy systems, Energy communities
Abstract: This paper presents a multi-frequency extension of the Impedance–Matching control framework for nonlinear Wave Energy Converters based on a spectral-domain technique. Building on previous single-frequency formulations, the proposed approach identifies the impedance-matching transfer function over a representative frequency range of the wave excitation spectrum, enabling improved energy absorption under irregular sea conditions. Two feedback controllers are compared: a conventional reactive controller and a broadband controller tuned via spectral-domain identification. Numerical experiments on a nonlinear point absorber model demonstrate enhanced absorbed power and improved phase alignment between excitation force and velocity, validating the effectiveness of the proposed approach.

Keywords: Control and management of energy systems, Demand response, Big data and machine learning applied to smart cities
Abstract: Buildings account for approximately 40% of global energy consumption, and with the growing share of intermittent renewable energy sources, enabling demand-side flexibility, particularly in heating, ventilation and air conditioning systems, is essential for grid stability and energy efficiency. This paper presents a safe deep reinforcement learning-based control framework to optimize building space heating while enabling demand-side flexibility provision for power system operators. A deep deterministic policy gradient algorithm is used as the core deep reinforcement learning method, enabling the controller to learn an optimal heating strategy through interaction with the building thermal model while maintaining occupant comfort, minimizing energy cost, and providing flexibility. To address safety concerns with reinforcement learning, particularly regarding compliance with flexibility requests, we propose a real-time adaptive safety-filter to ensure that the system operates within predefined constraints during demand-side flexibility provision. The proposed real-time adaptive safety filter guarantees full compliance with flexibility requests from system operators and improves energy and cost efficiency — achieving up to 50% savings compared to a rule-based controller — while outperforming a standalone deep reinforcement learning-based controller in energy and cost metrics, with only a slight increase in comfort temperature violations.

Keywords: Control and management of energy systems, Distributed optimization and control for smart cities, Thermal systems modelling
Abstract: District heating networks are projected to play a major role in the decarbonisation of building energy use, but traditional architectures are subject to significant inefficiency due to heat losses. Splitting networks into multiple lower-temperature zones is a promising method of reducing heat losses, but introduces privacy and coordination concerns. We show that the problem of optimising a multi-zone district heating network can be cast as a multi-agent constraint-coupled optimisation problem with linear coupling constraints, and we propose a decentralised algorithm based on the dual decomposition technique for its solution. Numerical examples show that that our decentralised approach recovers the same solution as optimising centrally, and that the problem exhibits both the static and time-varying turnpike properties.

Keywords: Control and management of energy systems
Abstract: This paper presents a Stochastic Tube-Based Economic Model Predictive Control (ST-EMPC) framework for microgrid energy management under Gaussian demand uncertainty. Existing tube-MPC methods rely on fixed or simplified uncertainty bounds and do not widely exploit forecasting-based uncertainty information. Therefore, we propose a controller that incorporates Gaussian disturbances through chance constraints driven by a forecast-dependent standard deviation, which enables adaptive constraint tightening and reduces robust MPC conservatism while maintaining feasibility. The framework is implemented in a grid-connected and multi-energy microgrid. Simulation results show full constraint feasibility, millisecond-level solve times, improved economic performance with only a 1.7% increase in cost compared to the oracle case, where the demand is fully known, and reliable operation under uncertainty, demonstrating the effectiveness and practical scalability of the proposed approach.

Keywords: Control and management of energy systems, Solar energy
Abstract: Modern Photovoltaic (PV) plants operate in highly dynamic and uncertain environments where irradiance, temperature, and load can change rapidly. Most maximum power point tracking (MPPT) algorithms are tuned for average conditions and provide little insight into risk or safety, which is increasingly critical for reliable grid-connected PV. In this paper, we formulate MPPT in dynamic environments as a risk aware control problem and propose a hybrid architecture that couples a Conditional Value-at-Risk (CVaR) model predictive layer with shielded reinforcement learning policy (RL) for fast and safe tracking. The predictive layer plans risk sensitive reference trajectories while the shielded RL policy executes these commands at converter timescales under hard voltage-current constraints. We validate the approach on a PV array and DC-DC converter model under diverse irradiance and partial shading scenarios. Results demonstrate superior performance against classical MPPT algorithms and learning based baselines.

Keywords: Machine learning and artificial intelligence in chemical process control, Industrial applications of chemical process control, Advanced process control
Abstract: In this work we present an efficient and practically implementable approach for the application of reinforcement learning (RL)-based control in chemical process systems. This is an area that has yet to widely adopt RL-based control largely due to inherent challenges in trusting RL algorithms and the time-consuming process of training reliable agents. To address these challenges we leverage a class of RL algorithms termed Y-wise Affine Neural Network (YANN)- RL, which we have developed in our prior work (Braniff and Tian, 2025a). By strategically initializing actor and critic networks YANN-RL algorithms provide confident and interpretable starting points within control schemes. We apply this RL-based control approach to three different process engineering case studies publically available on the PC-Gym library (Bloor et al., 2026): (i) a continuous stirred tank reactor (CSTR), (ii) a four-tank system, and (iii) a multistage extraction column. Our approach is compared to several popular RL algorithms (PPO, SAC, DDPG, and TD3) and is benchmarked against nonlinear model predictive control (NMPC). These case studies demonstrate that YANN-RL can greatly reduce the training time and data needed, can be deployed with confidence for chemical process systems, and can approach the performance of NMPC without the knowledge of a full nonlinear model.

Keywords: Machine learning and artificial intelligence in chemical process control, Advanced process control
Abstract: Achieving reliable and adaptive control of VSA processes for CO₂ capture remains challenging due to nonlinear cyclic transients, purity–recovery trade-offs, and sensitivity to variations in inlet composition. This work investigates a deep reinforcement learning controller that manipulates adsorption and evacuation durations to track recovery targets while enforcing product purity. Using a high-fidelity VSA simulator for training, the learned policy exhibits stable steady-state operation, effective tracking, and robustness to disturbances, demonstrating the potential of DRL for autonomous cycle-time control.

Keywords: Model-predictive and optimization-based control in chemical processes, Machine learning and artificial intelligence in chemical process control, Real-time optimization and control in chemical processes
Abstract: Multi-timescale systems present a fundamental challenge, where fast operational decisions must coexist with long-horizon sustainability targets. In this work, we proposed a new hierarchical control framework via Model Predictive Control (MPC) and Reinforcement Learning (RL) to separate decision-making on two distinct timescales. The high-level MPC optimizes long-horizon setpoints at the slow dynamic and on a fast timescale, a low-level pretrained RL agent tracks these setpoints in real time to maximize short-term objectives. RL is introduced to learn nonlinear control policies, without relying on model linearizations or requiring the heavy online computation from solving repeated optimal control problems. The framework is applied to a Battery Energy Storage System (BESS) operating in frequency regulation markets to balance fast profit opportunities (seconds) and slow battery degradation (weeks to months). The design employs a degradation-aware RL agent trained offline to generate safe long-horizon setpoints, and a degradation-unaware agent fine-tuned from it for fast runtime setpoint tracking. Compared to MPC baselines, the proposed approach successfully extends battery lifetime by 84% and increases operational profit by 34%.

Keywords: Control and optimization of supply chains in chemical processes
Abstract: Supply chains are interconnected networks of processes and operations producing and delivering high-value products. These chains are increasingly subjected to structural changes from the energy transition and other external factors. To address this, this work develops mixed-integer programming formulations to identify optimal reconfigurations that preserve industrial operations and profitability. We propose products and spatial neighborhoods to restrict the feasible search space and enable fast heuristic solutions. Furthermore, this restriction combines structural and product-based information, thus allowing to explore and define multiple reconfiguration scenarios. We demonstrate the approach using an agricultural waste case study, showing its ability to quickly produce good quality solutions.

Keywords: Control and optimization for sustainability and energy systems, Machine learning and artificial intelligence in chemical process control, Energy market
Abstract: The scheduling of energy storage can enable grid operators to promote stability, and process operators to benefit from the variability of modern electricity markets. Decision trees provide a paradigm to learn parsimonious optimal control policies while maintaining interpretability. In this work, we present a decision tree strategy to learn battery energy storage electricity trading policies from the solution to offline open-loop optimal control problems. We explore supervised and direct economic training approaches, finding that the learned policies can provide effective energy trading strategies even when compared to the open-loop policy with perfect market foresight. We find that the resulting tree policies are intuitive in terms of directionality and magnitude of their trading decisions. Our work provides a promising initial step to balance performance with interpretability in electricity arbitrage.

Keywords: Advanced process control
Abstract: Industrial injection molding operates in a complex and dynamic environment. To provide interpretable and transferrable monitoring, we propose the Physics-to-Semantics Multi-Agent System (PIMAS). This framework transforms high-frequency sensor data and process parameters into semantic text for intelligent monitoring. PIMAS introduces a two-stage adaptive mechanism. First, a perception agent converts raw sensor streams into semantic descriptors. It employs a self-calibrating mechanism to ensure adaptability across different machines and materials. Second, the system filters industrial noise using a multi-level tolerance rule. A Retrieval-Augmented Generation (RAG) diagnostic agent then identifies root causes. It retrieves universal physical principles from specialized knowledge bases, which avoids reliance on specific dataset patterns. Experiments on real-world datasets demonstrate that PIMAS maintains high diagnostic accuracy, even when transferred between different process settings. The proposed solution is flexible, explainable, and deployment-ready.

Keywords: AI and ML for environmental systems, Air quality modeling and control, Real time monitoring and control of environmental systems
Abstract: We propose a novel multi-modal adversarial framework for methane detection and quantification, fusing Airborne Visible InfraRed Imaging Spectrometer - Next Generation (AVIRIS-NG) with WorldView-3 satellite data. Our architecture utilizes parallel convolutional-transformer encoders linked by a bi-directional Cross-Attention Fusion module. To ensure physical consistency, we enforce radiative transfer constraints and a Fractional-Order Total Variation (FOTV) regularizer. The model generates methane enhancement maps, enabling flux estimation via Integrated Mass Enhancement (IME) and interpolated ERA5 wind fields. Qualitative Explainable AI (XAI) analysis confirms the model learns robust spectral-spatial features, overcoming single-sensor limitations. The proposed method is quantitatively compared with existing State of the art models for methane detection.

Keywords: Planning and management in environmental systems under deep uncertainty, Control of large-scale environmental systems, Participatory decision making in environmental systems
Abstract: Traditional centralized energy models often ignore national strategic interests and data privacy, assuming unrealistic perfect cooperation. To address this limitation, we propose a distributed multi-agent framework for the Southern African Power Pool, modeling coordination as a Distributed Constraint Optimization Problem (DCOP). Since standard DCOP solvers fail due to the combinatorial complexity of large-scale networks, we introduce our negotiation algorithm. Agents iteratively exchange bids based on marginal costs, overcoming these computational limits and preserving privacy. Simulations show rapid convergence that closely matches the centralized cost-optimal benchmark, validating the framework for realistic, distributed energy planning.

Keywords: Water resource system modeling and control, Modeling and estimation in agriculture, Optimal control and operation of environment systems
Abstract: Under the imperative to extract increasingly higher agricultural gains out of finite water resources, serious efforts are being put in to increase the productivity of water use in agricultural systems. In this paper, we develop a mathematical framework to investigate the tradeoff between productivity and equity in agricultural water allocation and present simulations inspired by a real-world irrigation system. We consider the effects of three different productivity enhancement initiatives: 1) Productivity enhancement through water allocation, 2) On-farm productivity enhancement, and 3) Enhancement in maximum crop yield. Our results indicate a clear tradeoff between equity and productivity in water allocation. However, on-farm productivity enhancements can increase overall productivity without any compromise on equity. Finally, while enhancement in maximum crop yield does enhance the total yield, it does not significantly affect either of the productivity or equity objectives. Thus, the potential of productivity to reconcile equity in agricultural water systems is different for different interventions and care must be exercised in promoting productivity gains as a single solution to uplift agricultural systems.

Keywords: Optimal control and operation of environment systems, Biological networks inference and modelling, Modeling and estimation in agriculture
Abstract: This paper addresses the optimization of nitrate fertilization by jointly considering the agronomic benefit of nitrate uptake and the economic and environmental costs associated with fertilizer use. Root absorption capacity of nitrate is known to evolve dynamically after fertilization, exhibiting a transient peak followed by a progressive decline, and fertilization strategies should account for this temporal behavior. We formulate the problem as an optimal control problem for a linear soil–plant model that incorporates nitrate dynamics in the soil, transporter activation, and feedback regulation exerted by amino acid accumulation. The model, obtained by linearizing a mechanistic nonlinear system, fits available experimental data and enables an explicit analytical solution of the optimal control law. Using a Pontryagin-based approach, the optimal strategy is determined in a single backward integration of the costate equation, and the fertilization interval is obtained directly from a threshold condition on the costate. Numerical simulations confirm that the optimal strategy consists of a single fertilization window and show that this strategy remains effective when applied to more detailed nonlinear models.

Keywords: Real time monitoring and control of environmental systems, AI and ML for environmental systems, Modeling and identification of environmental systems
Abstract: This paper presents an integrated framework for autonomous water quality monitoring using an autonomous surface vehicle (ASV). We address the challenge of inefficient coverage in dynamic aquatic environments through a genetic algorithm (GA) path planning approach that achieves 48.40% reduction in path length compared to the conventional methods. The framework combines an optimized coverage planning with robust nonlinear proportional-integral-derivative (PID) control and a gaussian process (GP)-based water quality estimation, validated using ROS 2/Gazebo simulations parameterized with real measurements of water quality data from Heron lake (France). The results demonstrate improved path tracking with an index of 0.41m root mean square error (RMSE) and effective spatiotemporal monitoring of key water parameters including dissolved oxygen(DO), pH, and temperature.

Keywords: Dynamics and control of biologically motivated nonlinear systems, Biological networks inference and modelling, Synthetic biology
Abstract: Achieving robustness to multi-parametric perturbations where all parameters can change at the same time is challenging because the controller would also face the same disturbance as the plant. For nonlinear positive feedback, an important mechanism for cell fate determination in biomolecular contexts, quantitative aspects of robustness to such perturbations are generally unclear. Here we used mathematical methods of control and dynamical systems, interval analysis, and a benchmark model of a bistable biomolecular positive feedback circuit to address this. We confirmed that such perturbations can change the qualitative behaviour of the system extinguishing bistability. We obtained a quantitative relation between the relative variations in the stable and unstable steady-states in terms of the relative changes in parameters. We showed how the deviation in the trajectories near the unstable steady-state due to such perturbations could diverge almost exponentially after an initial transient, which could have a significant impact on the bistable switching dynamics. We found that the size of the eigenvalue for the unstable steady-state was greater than that for the stable steady-state, and proved this for certain parameters using a rigorous numerical construction. We identified a trade-off between enhancing the parameter space of bistability and increased sensitivity in the bistable dynamics due to parametric perturbations. We obtained rigorous bounds on the entire transient response for these perturbations. These results provide a quantitative insight into the robustness of a bistable biomolecular positive feedback circuit to multi-parametric perturbations.

Keywords: Dynamics and control of biologically motivated nonlinear systems, Synthetic biology
Abstract: We present a compositional framework for verifying forward invariance of nonlinear dynamical systems with a cascaded structure, where variables are grouped into layers and each variable may be influenced by all preceding layers and the immediately following one. Our framework relies on barrier functions and exploits the cascade structure of the system to locally construct barrier functions for each layer, which are then combined into a non-smooth barrier function for the overall system. To show the efficacy of our framework, we develop an algorithmic framework to find barrier functions based on genetic algorithms and algebraic decompositions. On a set of experiments, including a non-linear 13-dimensional dynamical system, we demonstrate how our framework enables scalable and rigorous forward invariance analysis of dynamical systems with substantially improved performance compared to state-of-the-art methods.

Keywords: Dynamics and control of biologically motivated nonlinear systems, Synthetic biology, Biomedical system modeling, identification, and simulation
Abstract: We investigate the onset of a Hopf bifurcation in a minimal biologically plausible model motif of two genes with three types of interconnections: activation, repression and self-activation. We show that the basal activation rate acts as a bifurcation parameter, and we provide the exact analytic value of this parameter, at which the Hopf bifurcation occurs. Moreover, due to the fact that near the bifurcation point the covariance matrix of the linearized system diverges, we show that the equilibrium point starts to show fluctuations even before the bifurcation occurs. These fluctuations can be considered as early warning signals announcing that a Hopf bifurcation is going to occur. The proposed mathematical framework was developed analytically and illustrated through numerical simulations. Ultimately, we consider that these results may be of interest in the context of pre-disease detection.

Keywords: Modelling, parameter identification and state estimation in biosystems, Biological networks inference and modelling, Dynamics and control of biologically motivated nonlinear systems
Abstract: Gene activation cascades can function as molecular timers, although timing precision is limited by the inherent stochasticity of gene expression. Timing precision is typically quantified by the statistics of the First Passage Time (FPT), the time at which a downstream protein first reaches a functional threshold. While analytical results exist for single-gene systems and for multi-gene cascades with switch-like activation, the impact of graded activation on timing precision has been studied only numerically and lacks a complete analytical description. We address this gap by analyzing a two-gene cascade using a burst-dilution hybrid stochastic model. By defining a piecewise-linear activation function with a tunable activation threshold and slope, we derive exact solutions for the statistical moment dynamics of the FPT, avoiding the closure problems associated with nonlinear activation functions. For identical genes and a fixed maximum production rate, analytical approximations and simulations show that the optimal dose-response shifts from the linear-saturated limit at short timescales to an interior optimum with non-zero threshold and finite slope as the mean FPT increases. Furthermore, the optimal threshold increases monotonically with the mean FPT, while the optimal slope varies non-monotonically. These results establish an analytically tractable framework linking the shape of a graded activation function to timing precision in gene regulatory cascades.

Keywords: Synthetic biology, Modelling and control of microbial communities, Dynamics and control of biologically motivated nonlinear systems
Abstract: We investigate multicellular sender–receiver systems embedded in hydrogel beads, where diffusible signals mediate interactions among heterogeneous cells. Such systems are modeled by PDE–ODE couplings that combine three-dimensional diffusion with nonlinear intracellular dynamics, making analysis and simulation challenging. We show that the diffusion dynamics converges exponentially to a quasi-steady spatial profile and use singular perturbation theory to reduce the model to a finite-dimensional multi-agent network. A closed-form communication matrix derived from the spherical Green’s function captures the effective sender-receiver coupling. Numerical results show the reduced model closely matches the full dynamics while enabling scalable simulation of large cell populations.

Keywords: Agricultural robotics, Computer vision in agriculture, Sensing and perception in agriculture
Abstract: This article presents a visual–inertial system integrated into a surveying robot to monitor weed density in vineyards. The system combines stereo camera, IMU, and GPS data to evaluate weed control by comparing green-pixel regions (GPR) before and after operations. A two-module architecture detects and tracks GPR within regions of interest using HSV features, texture patterns, and depth data, while geolocating frames through interpolation. Results show effective performance, producing real-world GPR maps. Statistical analysis confirms significant differences between pre- and post-operation data, with overall reduction of 55.6% in intra-row GPR, supporting the system’s applicability for automated viticulture weed management.

Keywords: Autonomous mobile robots, Control architectures in automotive control, Digital twins and IoT for aerospace systems control and monitoring
Abstract: This paper presents a small-scale hybrid test-bed, MASCOT, developed for testing and validation of the control algorithms for multi-agent systems on multi-robot systems. Distributed control algorithms proposed in the literature use simplified models for agent dynamics like single or double integrator dynamics. MASCOT facilitates testing of such algorithms on real systems through simulations, physical experiments, and hybrid experiments with simplified user interface. MASCOT is developed using Robot Operating Systems 2 (ROS2) and Gazebo but provides user interface so that control engineers do not need the exposure to ROS2 or Gazebo. Currently, MASCOT uses Crazyflie 2.1 and Loco-positioning System for experimental part. Apart from distributed control algorithms, MASCOT also provides haptic interface for human-on-the-loop control strategies for multi-agent systems through bilateral teleoperation. The performance of the testbed is analyzed by implementing linear control laws such as leaderless consensus, leader-follower consensus, bearing based formation and non-linear control law for min-max time consensus. This work is published as an open-source ROS package under MIT license at https://github.com/IITDhHANS/mascotV2.git

Keywords: Autonomous mobile robots, Intelligent transportation systems, Automatic control, optimization, real-time operations in transportation
Abstract: High local density brings large swarms to a standstill by shrinking safety-feasible motion to near zero, so decentralized planners stall and form persistent deadlocks. We presents a density-regulated control framework that integrates macroscopic traffic regulation with agent-level safety control. A density-guided reference velocity respects regional capacity limits and modulates inflow so agents slow down or reroute before saturation. A high-order control barrier function controller tracks this reference at the acceleration level, enforces second-order dynamics, and guarantees collision avoidance. Experiments show deadlock removal and real--time performance at scale.

Keywords: Autonomous mobile robots
Abstract: Stable current collection in high-speed rail is critically dependent on the pantograph’s ability to accurately follow the overhead catenary wire. At high velocities, the wire exhibits rapid vertical displacement that the whole mechanism of the passive pantograph systems has difficulty responding to, leading to contact force fluctuation. To address this issue, this paper presents an active control framework for adjusting the pantograph head displacement to reduce the fluctuation of contact force. A multi-body dynamic model of the pantograph head is developed to simulate the interaction dynamics. Subsequently, a PI-Lead controller is designed to actively regulate the pantograph head’s position with a fast-dynamic response. Simulation results validate the proposed controller, showing it effectively minimizes the contact force fluctuation by compensating for rapid wire movement, thereby maintaining a stable contact.

Keywords: Autonomous mobile robots, Automatic control, optimization, real-time operations in transportation, Intelligent transportation systems
Abstract: The need for explainability to improve resilience and trustworthiness is a hot topic in various fields supported by autonomous decision-making systems. This motivates the development of systematic methods that are able to explain decisions to users in real-time, especially if the complex system is interacts with human beings. Improving the efficiency of the explainable formulation requires a rigorous definition of the target audience, because explainability level differs for everyday users and experts. This difference leads to the challenges related to defining explainability levels and selecting appropriate approximation structures. This paper proposes a methodology by which the explainability formulation at multiple levels can be effectively carried out. The methodology is presented based on a case study in which autonomous mobile robots perform packaging mission in a warehouse logistic system. The goal of the formulation is to explain healthy or faulty operation of the robots for professional, operator, and general users.

Keywords: Automatic control, optimization, real-time operations in transportation, Aerospace mission control and operations, Multi-vehicle systems
Abstract: Transporting loads using multiple drones offers many advantages, including improved delivery performance and higher system resilience, yet achieving coordinated and energy-efficient control remains challenging. This study proposes a formation transition strategy for cooperative multi-drone payload transport that integrates trajectory control, collision avoidance, and dynamic formation optimization. A generalized dynamic model and a sliding-mode controller ensure robust tracking, while a virtual leader coordinates drones during transitions. Energy efficiency is enhanced using a Late Acceptance Hill Climbing algorithm. Simulations show reduced energy use and more balanced thrust distribution, achieving a 45.6% improvement.

Keywords: Intelligent transportation systems, Cooperative navigation, Autonomous mobile robots
Abstract: This paper proposes a trust-aware distributed observer for vehicle platoons that maintains resilient state estimation under cyberattacks. A behavioral divergence metric evaluates the reliability of shared data, forming a dynamic neighbor set used to adapt observer's weighting gains. Stability conditions are derived via Lyapunov analysis. Simulations under bogus, replay, and DoS attacks demonstrate robust performance and stable platoon behavior.

Keywords: AI and embodied-AI in marine systems, Modelling, identification and control in marine systems, Simulation and digital-twin in marine systems
Abstract: This paper explores a data-driven motion-based wave estimation system from the measured response of a dynamically positioned drilling vessel. Sea state estimation models using neural networks were trained exclusively with simulated data coming from the vessel’s dynamic model subjected to metocean conditions collected from reanalysis datasets. Inference performance from in-service motion records was assessed through direct comparison with reanalysis data and via an indirect evaluation scheme by analyzing the response of the real drilling vessel and the corresponding simulation driven by the estimated sea states. Results indicate the potential of the approach to provide reliable wave estimates to DP systems.