Keywords: Adaptive control, Automotive systems, Reinforcement learning
Abstract: As Hybrid Electric Vehicles (HEVs) gain traction in heavy-duty trucks, adaptive energy management is critical for minimizing fuel consumption and maintaining battery charge over long operations. To address this, we present a reinforcement learning (RL) framework for engine control in Series HEVs, based on Soft Actor-Critic (SAC) framed as a sequential decision-making problem. Specifically, standard Feedforward networks (FFNs) in the actor and critic are replaced with temporal models, Gated Recurrent Units (GRUs), and Decision Transformers (DTs), to better capture sequential dependencies. Through comprehensive ablation studies, we identify two high-performing and well-generalized sequence-aware agent configurations: a DT actor with a GRU critic (DT-GRU), and a GRU actor with a GRU critic (GRUGRU). Validation compared these SAC agents against Dynamic Programming (DP) across two training setups: one trained on a single standard cycle (HFET), and another using a wider range of initial SOCs, drive durations, and power demands. On the training cycle (HFET), the DT-GRU agent achieved fuel performance within 1.8% of DP, outperforming the GRU-GRU agent, and the FFN-FFN agent at 3.16% and 3.43%, respectively. The unseen highway test cycles, one with significantly higher power demand (US06), while another with extended operation time (HHDDT), compared to the training cycle, allowed us to assess robustness. Both sequence-aware agents showed better generalization than the FFN-FFN agent across these challenging conditions. Notably, agents trained with broader initial conditions not only maintained strong fuel economy but also satisfied critical final constraints.

Keywords: Adaptive control, Biological systems, Modeling

Keywords: Adaptive control, Control applications, Constrained control
Abstract: Maximizing advertiser value under budget constraints is a central challenge in programmatic advertising. This paper presents a unified framework that addresses optimal bidding, budget allocation, and real-time spend control in campaigns with multiple ad groups, each subject to individual spend constraints. We first derive the structure of an optimal bidding policy that maximizes expected return on investment in first-price auctions. Building on this, we introduce an algorithm for allocating campaign budgets optimally across ad groups in real time, ensuring compliance with both campaign-level and ad group-level constraints. Finally, we propose a feedback-control system that dynamically adjusts bidding parameters to achieve and maintain optimal spend distribution under changing market conditions. Simulations demonstrate convergence to optimal allocations and robust behavior in both constrained and unconstrained scenarios. By designing the system with dependence only on practically estimable quantities, this work bridges the gap between theoretically optimal bidding strategies, and real-world real-time bidding applicability.

Keywords: Adaptive control, Discrete event systems, Stability of nonlinear systems
Abstract: We study a general class of nonlinear discrete-time systems containing disturbances in the dynamics and in the available output measurements, and unknown control directions. Under suitable conditions on the coefficient matrices in the linear portion of the dynamics and on a coefficient matrix in the output measurement, and under an observability condition, we provide a new method to identify the unknown control direction in finite time. Our method is amenable to systems whose linear portions contain uncertain coefficients. We illustrate our method using a discrete-time chain of integrators.

Keywords: Adaptive control, Flight control
Abstract: Many robotic systems are underactuated, meaning not all degrees of freedom can be directly controlled due to lack of actuators, input constraints, or state-dependent actuation. This property, compounded by modeling uncertainties and disturbances, complicates the control design process for trajectory tracking. In this work, we propose an adaptive control architecture for uncertain, nonlinear, underactuated systems with input constraints. Leveraging time-scale separation, we construct a reduced-order model where fast dynamics provide virtual inputs to the slower subsystem and use dynamic control allocation to select the optimal control inputs given the non-affine dynamics. To handle uncertainty, we introduce a state predictor-based adaptive law, and through singular perturbation theory and Lyapunov analysis, we prove stability and bounded tracking of reference trajectories. The proposed method is validated on a vertical takeoff and landing (VTOL) quadplane with nonlinear, state-dependent actuation, demonstrating its utility as a unified controller across various flight regimes, including cruise, landing transition, and hover.

Keywords: Adaptive control, Game theory, Optimization
Abstract: The nested Extremum Seeking (nES) algorithm is a model-free optimization method for tuning system parameters across multiple objective functions using bounded extremum seeking. Using weak-limit averaging and singular perturbation techniques, we establish a rigorous stability result proving semiglobal practical convergence to a Nash equilibrium. These results suggest that nES provides a principled approach for feedback-based tuning of coupled objectives, including settings which admit a game-theoretic interpretation.

Keywords: Adaptive control, Identification for control, Networked control systems
Abstract: Networked multi-agent dynamical systems have been used to model how individual opinions evolve over time due to the opinions of other agents in the network. Particularly, such a model has been used to study how a planning agent can be used to steer opinions in a desired direction through repeated, budgeted interventions. In this paper, we consider the problem where individuals’ susceptibilities to external influences are unknown. We propose an online algorithm that alternates between estimating this susceptibility parameter, and using the current estimate to drive the opinion to a desired target. We provide conditions that guarantee stability and convergence to the desired target opinion when the planning agent faces budgetary or temporal constraints. Our analysis shows that the key advantage of estimating the susceptibility parameter is that it helps achieve near-optimal convergence to the target opinion given a finite amount of intervention rounds, and, for a given intervention budget, quantifies how close the opinion can get to the desired target.

Keywords: Adaptive control, Identification for control, Neural networks
Abstract: This paper presents the first result that enables the use of a concurrent learning (CL) adaptation law for the weights of all the layers of the DNN-based controller applied to a class of second-order control-affine systems. The developed CL-based adaptation achieves convergence of the DNN’s parameter estimates to a neighborhood of their ideal values, provided the DNN’s Jacobian satisfies a finite excitation condition. A Lyapunov-based stability analysis is conducted to ensure convergence of the tracking error, weight estimation errors, and observer errors to a neighborhood of the origin.

Keywords: Adaptive control, Intelligent systems, Reinforcement learning
Abstract: This paper extends the single network adaptive critic structure, initially developed for near-optimal control of discrete-time systems with offline training, to an online continuous-time setting. A novel gradient-descent-based training law is proposed, and its exponential convergence is established under the persistency of excitation assumption. The stability of the closed-loop system is analyzed using Lyapunov’s direct method, and the uniformly ultimately bounded nature of the closed-loop tracking error is proved. Numerical simulations on both tracking and regulation tasks demonstrate the effectiveness of the proposed approach.

Keywords: Adaptive control, Optimization, Robust adaptive control
Abstract: This paper studies the bearing-only Fermat–Weber Location Problem (FWLP) for Euler–Lagrange systems. The objective is to design bearing-only control laws that guide an autonomous robot to the point that minimizes the weighted sum of Euclidean distances to a set of beacons, known as the Fermat–Weber point. In contrast to previous studies that focus on simplified models such as single- or double-integrator dynamics, this work considers Euler–Lagrange systems, which offer a more realistic representation of physical robot dynamics. The physical parameters of the robot are considered unknown. For stationary beacons, we develop a bearing-only control law using an adaptive backstepping method to steer the robot to the Fermat–Weber point. We then extend the analysis to a case in which the beacons move with a constant velocity, and the dynamics of the robot is affected by disturbances. To address this scenario, we propose a novel robust adaptive bearing-only control law that allows the robot to track the moving Fermat–Weber point. Both proposed control laws guarantee asymptotic convergence, and simulation results are provided to validate the theoretical findings.

Keywords: Adaptive control, Neural networks, Stability of nonlinear systems
Abstract: This paper presents a novel approach to solving the indirect influence problem in networked systems, in which cooperative nodes must regulate a target node with uncertain dynamics to follow a desired trajectory. We leverage the message-passing structure of a graph neural network (GNN), allowing nodes to collectively learn the unknown target dynamics in real time. We develop a novel GNN-based backstepping control strategy with formal stability guarantees derived from a Lyapunov-based analysis. Numerical simulations are included to demonstrate the performance of the developed controller.

Keywords: Adaptive control, Optimization, Cooperative control
Abstract: This paper studies the safe and resilient control of Connected and Automated Vehicles (CAVs) operating in mixed traffic environments where they must interact with Human-Driven Vehicles (HDVs) under uncertain dynamics and exponentially unbounded false data injection (EU-FDI) attacks. These attacks pose serious threats to safety-critical applications. While resilient control strategies can mitigate adversarial effects, they often overlook collision avoidance requirements. Conversely, safety-critical approaches tend to assume nominal operating conditions and lack resilience to adversarial inputs. To address these challenges, we propose an event-driven safe and resilient (EDSR) control framework that integrates event-driven Control Barrier Functions (CBFs) and Control Lyapunov Functions (CLFs) with adaptive attack-resilient control. The framework further incorporates data-driven estimation of HDV behaviors to ensure safety and resilience against EU-FDI attacks. Specifically, we focus on the lane-changing maneuver of CAVs in the presence of unpredictable HDVs and EU-FDI attacks on acceleration inputs. The event-driven approach reduces computational load while maintaining real-time safety guarantees. Simulation results, including comparisons with conventional safety-critical control methods that lack resilience, validate the effectiveness and robustness of the proposed EDSR framework in achieving collision-free maneuvers, stable velocity regulation, and resilient operation under adversarial conditions.

Keywords: Adaptive systems, Uncertain systems, Lyapunov methods
Abstract: The presence of system anomalies, such as modeling uncertainties and unknown actuator degradation, weakens the performance of controlled systems when not properly compensated. Thus, in this paper, for improved closed-loop performance, a novel high-performance model reference adaptive controller, along with an adaptive control allocation method, is designed for uncertain over-actuated nonlinear dynamical systems in the presence of unknown actuator degradation. A nonlinear reference model is used to track the considered dynamical system, since standard linear reference models limit the closed-loop performance of the nonlinear system. A Lyapunov stability analysis is then used to show the asymptotic convergence of the tracking error between the closed-loop system and the nonlinear reference model. Simulations are also conducted to show the benefits of the proposed method.

Keywords: Adaptive control, Optimization algorithms, Estimation
Abstract: Extremum-seeking control (ESC) is widely used to optimize performance when the system dynamics are uncertain. However, sensitivity to sensor noise is a crucial issue in ESC implementation due to the use of high-pass filters or gradient estimators. To reduce the sensitivity of ESC to noise, this paper investigates the use of the recently developed adaptive input and state estimation (AISE) technique for numerical differentiation. In particular, this paper develops extremum-seeking control with adaptive input and state estimation (ESC/AISE), where AISE replaces the high-pass filter of ESC to improve performance under sensor noise. The effectiveness of ESC/AISE is illustrated via numerical examples.

Keywords: Adaptive control, Stochastic optimal control, Automotive control
Abstract: Accurate predictive control requires not only reliable parameter estimates but also correct models of system dynamics. In many real world applications, multiple candidate models can describe the system equally well, and conventional controllers typically rely on passive adaptation that fails to actively reduce such model uncertainty. This paper presents a Dual Model Predictive Path Integral (dual-MPPI) controller for online model discrimination, enabling autonomous systems to both act and learn under competing model hypotheses. By combining sampling-based rollout evaluation with Bayesian updates of model probabilities and parameter beliefs, the proposed method balances exploration and exploitation during control. Effectiveness is demonstrated on a simulated off-road vehicle navigating terrains represented by two distinct tire-terrain interaction models. Compared against a certainty-equivalent MPPI baseline, dual-MPPI proactively issues exploratory actions that accelerate model discrimination while maintaining safety, resulting in faster convergence of model probabilities and improved vehicle stability in dynamic maneuvers. These result highlight the potential of dual-MPPI to address both parametric and model uncertainty in safety-critical autonomous systems.

Keywords: Adaptive control, Switched systems, Observers for nonlinear systems
Abstract: This letter considers an autonomous agent that intermittently acquires state measurements to maintain trajectory tracking performance. The objective is to minimize sensing needs using extended periods of sensor-denied operation. A Lyapunov-based adaptive switched systems approach is developed, where the agent uses the intermittently acquired state measurements to learn the system model. The learned system models are then used during sensor-denied intervals to extend their length while maintaining tracking performance. The design uses a modeling error-dependent bound on the duration of the sensor-denied intervals to progressively reduce sensing needs as the modeling error decreases. A Lyapunov-based analysis is used to show the stability of the closed-loop system, and the effectiveness of the developed technique is verified through a simulation study.

Keywords: Adaptive control, Uncertain systems, Time-varying systems
Abstract: We provide a new method to compute signs of unknown control directions for a general class of time-varying nonlinear systems that contain unknown disturbances in the available output and in the dynamics. Our relation between the output and a matrix factor in the control effectiveness matrix guarantees that the sign can be computed in an arbitrarily small time. Our method yields bounds on the allowable uncertainty in the coefficient matrices. We illustrate our method using a nonlinear perturbed generalization of a chain of integrators.

Keywords: Adaptive systems, Direct adaptive control, Uncertain systems
Abstract: In this work, we propose a discrete-time adaptive control architecture for uncertain fully-actuated systems in the presence of unknown control effectiveness, modeled by a diagonal matrix that multiplies the control input vector in the dynamical model. To establish boundedness of the closed-loop system and convergence of the asymptotic tracking error, a Lyapunov stability analysis is conducted. Moreover, simulation results demonstrate the efficacy of the proposed method.

Keywords: Estimation, Adaptive control, Neural networks
Abstract: We present a novel formulation of the Tobit Kalman Filter (TKF) for an inertial navigation problem with censored GNSS measurements. Our formulation utilizes the IMU measurement for the propagation step, and the last known GNSS position before censorship as a measurement for the update step. To calculate the innovation and the Kalman gain under censored measurements, we derive new formulas for the predicted measurement and the covariance terms using likelihood functions based on a statistical Tobit model. Therefore, TKF allows for state estimation despite the stagnant measurements. Additionally, due to the coupling between the navigation and control problems, censored measurements contribute to the increased uncertainty of the overall system dynamics. From the control perspective, the uncertainty can be handled by an adaptive control technique. Hence, we integrate a dynamic inversion adaptive controller to complete the system autonomy. We use fixed-wing airplane dynamics with a neural network (NN) as an adaptive element. The NN weights are updated online using a law derived from the Lyapunov theorem. The proposed approach is evaluated using a UAV flight simulator. The result confirms that this approach makes navigation and control in a censored measurement environment feasible.

Keywords: Estimation, Constrained control, Variational methods
Abstract: We propose a control barrier function (CBF) formulation for enforcing equality and inequality constraints in variational inference. The key idea is to define a barrier functional on the space of probability density functions that encode the desired constraints imposed on the variational density. By leveraging the Liouville equation, we establish a connection between the time derivative of the variational density and the particle drift, which enables the systematic construction of corresponding CBFs associated to the particle drift. Enforcing these CBFs gives rise to the safe particle flow and ensures that the variational density satisfies the original constraints imposed by the barrier functional. This formulation provides a principled and computationally tractable solution to constrained variational inference, with theoretical guarantees of constraint satisfaction. The effectiveness of the method is demonstrated through numerical simulations.

Keywords: Estimation, Distributed control, Autonomous robots
Abstract: Reliable localization in GPS-denied environments remains challenging for multi-robot systems. Although visual–inertial odometry (VIO) provides accurate motion estimation, it suffers from drift without global references. Ultra-Wideband (UWB) ranging offers complementary constraints, but existing UWB-aided VIO methods typically assume single-robot operation and known anchor locations. We propose a distributed collaborative visual–inertial–ranging odometry (DC-VIRO) framework that tightly fuses VIO and UWB across multiple robots. Anchor states are jointly estimated, and shared observations introduce additional geometric constraints through inter-robot communication. A right-invariant formulation preserves VIO unobservable directions, ensuring consistency. Simulations show improved localization accuracy, robustness, and distributed anchor self-calibration.

Keywords: Estimation, Filtering, Machine learning
Abstract: Target tracking using bearings-only information is a well-known challenge, primarily attributed to the system's poor observability and the highly nonlinear nature of the estimation process. In this paper, a deep learning framework based on the Gated Recurrent Unit (GRU) architecture is proposed to address this bearings-only tracking problem. Different from conventional nonlinear filters that rely on explicit system models, the proposed data-driven method learns to estimate the target state directly from sequential noisy bearing observations. A comprehensive comparative study is conducted to evaluate the proposed model against other prominent deep learning architectures, including the Transformer, Long Short-Term Memory (LSTM), and Convolution Neural Network (CNN), as well as traditional nonlinear filters. Simulation results demonstrate that the proposed GRU-based tracker provides the most accurate position estimates in a nominal scenario. In a challenging scenario with high uncertainty, the GRU model attains the best balance among position and velocity estimation accuracies and computation efficiency. This highlights the strong robustness and stability of the GRU, and its great potential for applications in modern passive tracking systems.

Keywords: Estimation, Kalman filtering, Delay systems
Abstract: This paper presents a Delayed Constrained Extended Kalman Filter (DC-EKF) that jointly addresses measurement delays and quaternion norm preservation. Existing constrained filters typically handle delays or norm constraints separately, degrading accuracy in multi-sensor fusion. The proposed method reformulates the delayed system as an augmented, delay-free model in which each quaternion state satisfies the unit-norm constraint. Multiple selector matrices are employed to decompose the multi-constraint optimization into independent components, yielding a closed-form Kalman gain via Lagrange multipliers. This approach avoids conventional partially constrained filters' partitioning and reassembly steps while maintaining computational efficiency. Simulation studies of rigid-body motion demonstrate that the DC-EKF achieves faster convergence, lower estimation variance, and stricter quaternion norm adherence compared to the EKF, pseudo-measurement EKF, and augmented constrained EKF. On average, settling times are reduced by 50--90% and steady-state errors by 70--90%, while quaternion norms remain within 10^{-6} of unity. These results highlight DC-EKF as an effective and practical solution for accurate pose estimation with delayed measurements.

Keywords: Estimation, Kalman filtering, Machine learning
Abstract: Popular Bayes filters often apply linearization techniques, such as Taylor expansion or stochastic linear regression, to enable the use of the Kalman filter structure, but this can lead to large errors in strongly nonlinear systems. The recently proposed NANO filter addresses this issue by interpreting the prediction and update steps of Bayesian filtering as two distinct optimization problems and solving them through moment matching and natural gradient descent, thereby avoiding model linearization errors. However, the natural gradient update in NANO can occasionally diverge because the posterior covariance in its iteration may lose positive definiteness. Our analysis shows that the posterior covariance is the sum of the inverse prior covariance and the expected Hessian of the log-likelihood function, and that the indefiniteness of the latter term is the root cause of update failure. To address this issue, we propose two remedies. The first approximates the log-likelihood Hessian using the Gauss–Newton method, representing it as the self-adjoint product of the Jacobian of the normalized measurement residual, which is guaranteed to be positive semi-definite. The second reformulates the covariance update as an exponential-form update of the Cholesky factor and reconstructs the covariance via its Gram matrix, which ensures positive definiteness. Experiments on three classical nonlinear systems demonstrate that the proposed NANO filter with guaranteed positive definiteness outperforms popular members of the Kalman filter family and original NANO filter.

Keywords: Estimation, Kalman filtering, Robust control
Abstract: This paper focuses on the linear quadratic control (LQC) design of systems corrupted by both stochastic noise and bounded noise simultaneously. When only one type of these noises is considered, the LQC strategy leads to either a stochastic or a robust controller. However, there is no LQC strategy that can simultaneously handle stochastic and bounded noises efficiently. This limits the scope where existing LQC strategies can be applied. In this work, we look into the LQC problem for discrete-time systems that have both stochastic and bounded noises in their dynamics. We develop a state estimation method for such systems by efficiently combining a Kalman filter and an ellipsoid set-membership filter. The developed estimator can recover the estimation optimality when the system is subject to both kinds of noise, the stochastic and the bounded. Leveraging the developed state estimation, we derive a robust state-feedback control law for the LQC problem. The control law derivation takes into account both stochastic and bounded-state estimation errors, so as to avoid over-conservativeness while sustaining stability in the control. In this way, the developed LQC strategy extends the range of scenarios where LQC can be applied, especially those of real-world control systems with diverse sensing which are subject to different kinds of noise. We present numerical simulations, and the results demonstrate the enhanced control performance with the proposed strategy.

Keywords: Estimation, Mechanical systems/robotics, Nonholonomic systems
Abstract: This work demonstrates simultaneous pose (position and orientation) and shape estimation for a free-floating, bioinspired multi-link robot with unactuated joints, link-mounted thrusters for control, and a single gyroscope per link, resulting in an underactuated, minimally sensed platform. Because the inter-link joint angles are constrained, translation and rotation of the multi-link system requires cyclic, reciprocating actuation of the thrusters, referred to as a gait. Through a proof-of-concept hardware experiment and offline analysis, we show that the robot's shape can be reliably estimated using an Unscented Kalman Filter augmented with Gaussian process residual models to compensate for non-zero-mean, non-Gaussian noise, while the pose exhibits drift expected from gyroscope integration in the absence of absolute position measurements. Experimental results demonstrate that a Gaussian process model trained on a multi-gait dataset (forward, backward, left, right, and turning) performs comparably to one trained exclusively on forward-gait data, revealing an overlap in the gait input space, which can be exploited to reduce per-gait training data requirements while enhancing the filter's generalizability across multiple gaits. Lastly, we introduce a heuristic derived from the observability Gramian to correlate joint angle estimate quality with gait periodicity and thruster inputs, highlighting how control affects estimation quality.

Keywords: Estimation, Neural networks, Pattern recognition and classification
Abstract: This paper tackles the problem of class conditional generation for spatial data represented as unordered sets of graph-structured nodes. Each node encodes spatial coordinates and activation intensities, and the goal is to model the conditional distribution over such sets, given a class label. We propose a Conditional Graph Variational Autoencoder (CGVAE) that learns to estimate class-aware latent representations and reconstruct node sets in a statistically grounded probabilistic framework. Our approach conditions both the encoder and decoder networks on class labels using Feature-wise Linear Modulation (FiLM) layers, enabling accurate posterior estimation while avoiding the need for per class decoders. To ensure stable and meaningful learning, we introduce a regularization strategy based on controlled information flow in the latent space. We employ a customized KL divergence loss which ensures that the model’s internal predictions remain close to a target distribution and a permutation-invariant reconstruction loss that compares both the node positions and intensities. This helps the generated data to closely match real data in a reliable way. Results on the MNIST superpixels demonstrate that the presented model is capable of generating diverse and label-consistent samples, and can support manipulations of geometric attributes such as slant, width, and height through latent space optimization. Further analysis reveals that the class identity is controlled by the FiLM layer, while geometric variation within each class is captured in the latent space, thus highlighting the model’s ability to separately estimate structured components of the data.

Keywords: Fuzzy systems, Estimation
Abstract: This work presents an observer based estimation method for the model uncertainties of super coiled polymer (SCP) actuators using an adaptive fuzzy logic (AFL) based approach. Specifically, an observer formulation based on fuzzy logic estimator is proposed for the uncertainties present in the nonlinear model of the SCP actuator. The proposed observer utilizes the known parts of the system model while employing AFL to estimate the unknown parameters and uncertainties. The design incorporates a self-learning fuzzy logic based term, where the control representative value matrix, as well as the means and variances of the membership functions, are dynamically updated. The stability and convergence of the overall system is ensured rigorously via Lyapunov type arguments. Simulation studies are presented to illustrate the performance and viability of the proposed method.

Keywords: Estimation, Optimization, Robotics
Abstract: A data-driven framework is proposed for online estimation of quadrotor motor efficiency via residual minimization. The problem is formulated as a constrained nonlinear optimization that minimizes trajectory residuals between measured flight data and predictions generated by a quadrotor dynamics model. A sliding-window strategy enables online estimation, and the optimization is efficiently solved using an iteratively reweighted least squares (IRLS) scheme combined with a primal-dual interior-point method, with inequality constraints enforced through a logarithmic barrier function. Robust z-score weighting is employed to reject outliers, which is particularly effective in motor clipping scenarios where the proposed estimator exhibits smaller spikes than an EKF baseline. Compared to traditional filter-based approaches, the batch-mode formulation allows selective inclusion of data segments via IRLS reweighting and hard-rejection. This structure is well-suited for online estimation and supports applications such as fault detection and isolation (FDI), health monitoring, and predictive maintenance in aerial robotic systems. Simulation results under various degradation scenarios demonstrate the accuracy and robustness of the proposed estimator.

Keywords: Estimation, Smart grid, Neural networks
Abstract: Baseline estimation is critical to Demand Response (DR) settlement in electricity markets, yet existing machine learning methods remain limited in predictive performance, while methodologies from causal inference and counterfactual prediction are still underutilized in this domain. We introduce a Generalized Synthetic Control Method that builds on the classical Synthetic Control Method (SCM) from econometrics. While SCM provides a powerful framework for counterfactual estimation, classical SCM remains a static estimator: it fits the treated unit as a combination of contemporaneous donor units and therefore ignores predictable temporal structure in the residual error. We develop a generalized SCM framework that transforms baseline estimation into a dynamic counterfactual prediction problem by augmenting the donor representation with exogenous features, lagged treated load, and selected lagged donor signals. This enriched representation allows the estimator to capture autoregressive dependence, delayed donor-response patterns, and error-correction effects beyond the scope of standard SCM. The framework further accommodates nonlinear predictors when linear weighting is inadequate, with the greatest benefit arising in limited-data settings. Experiments on the Ausgrid smart-meter dataset show consistent improvements over classical SCM and strong benchmark methods, with the dominant performance gains driven by dynamic augmentation.

Keywords: Robotics, Intelligent systems, Sensor fusion
Abstract: This paper presents a novel particle filter-based approach for real-time human intention prediction in human-robot interaction (HRI) scenarios. Unlike existing methods that rely on offline training or single-modality inputs, our framework fuses three complementary information sources—head orientation, hand orientation based on object affordances, and hand velocity vectors—to predict which object a human intends to interact with. The particle filter naturally accommodates dynamic scenes where objects can be added or removed without requiring retraining, addressing a critical limitation of current learning-based approaches. We validate our method using a custom dataset of 48 grasping trajectories across 16 experiments, achieving approximately 90% accuracy in intention prediction with processing times of 2.7ms per frame. Comparative evaluation against a Bayesian Network approach demonstrates comparable performance (80.6% vs. 83.1% accuracy) despite requiring no offline training. Furthermore, we demonstrate the system’s adaptability through simulated object addition experiments, where the particle filter maintains stable performance as the state space dynamically expands. Our approach offers a practical solution for real-time human intention prediction in collaborative robotics applications, eliminating the need for task-specific training while maintaining computational efficiency suitable for real-time control.

Keywords: Sensor networks, Optimization algorithms, Filtering
Abstract: This paper studies sensor placement when detection performance varies stochastically due to environmental factors over space and time and false alarms are present, but a filter is used to attenuate the effect. We introduce a unified model that couples detection and false alarms through an availability function, which captures how false alarms reduce effective sensing and filtering responses to the disturbance. Building on this model, we give a sufficient condition under which filtering improves detection. In addition, we derive a coverage-based lower bound on the void probability. Furthermore, we prove robustness guarantees showing that performance remains stable when detection probabilities are learned from limited data. We validate the approach with numerical studies using AIS vessel-traffic data and synthetic maritime scenarios. Together, these results provide theory and practical guidance for deploying sensors in dynamic, uncertain environments.

Keywords: Observers for nonlinear systems, H-infinity control, Spacecraft control
Abstract: This paper develops a robust estimation framework for cislunar navigation that embeds the Circular Restricted Three-Body Problem (CR3BP) dynamics and bearing-only optical measurements within a Linear Fractional Transformation (LFT) representation. A full-order mathcal{H}_infty observer is synthesized with explicit mathcal{L}_2 performance bounds. The formulation yields a nonlinear estimator that operates directly on the governing equations and avoids reliance on local linearizations. Dominant nonlinearities are expressed as structured real uncertainties, while measurement fidelity is represented through range-dependent weighting with Earth-Moon distances reconstructed from line-of-sight geometry. The sensing architecture assumes passive star-tracker-class optical instruments, eliminating the need for time-of-flight ranging or precision clocks. Simulations demonstrate bounded estimation errors and smooth position tracking over multiple orbital periods, with the largest deviations observed in the out-of-plane states, consistent with the stiffness of the vertical dynamics and the limitations of angle-only observability. Application to a Near Rectilinear Halo Orbit (NRHO) illustrates that the framework can achieve robust onboard navigation with bounded estimation errors using flight-representative sensors.

Keywords: Observers for nonlinear systems, Robust control, Learning
Abstract: This paper presents a data–driven approach for jointly learning a robust full–state observer and its robustness certificate for systems with unknown dynamics. Leveraging incremental input–to–state stability (δISS) notions, we jointly learn a δISS Lyapunov function that serves as the robustness certificate and prove practical convergence of the estimation error under standard fidelity assumptions on the learned mod- els. This renders the observer safety–compatible: they can be consumed by certificate–based safe controllers so that, when the controller tolerates bounded estimation error, the controller’s certificate remains valid under output feedback. We further extend the approach to interconnected systems via the small– gain theorem, yielding a distributed observer design framework. We validate the approach on a variety of nonlinear systems.

Keywords: Quantum information and control, Game theory
Abstract: In this tutorial, we explore the intersection of quantum control, game theory, and artificial intelligence deception, emphasizing their combined potential to enhance secure and strategic decision-making in complex systems. It examines how deceptive behavior can be understood and influenced within strategic frameworks, and outlines principles for designing resilient systems that leverage advantages in both quantum and learning-driven environments. While the discussion centers on specific examples, the foundational ideas are broadly applicable across a range of domains.

Keywords: Spacecraft control, Optimal control, Constrained control
Abstract: Soft landing on small celestial bodies (SCBs) poses unique challenges, as uncertainties in gravitational models and poorly characterized, dynamic environments require a high level of autonomy. Existing control approaches lack formal guarantees for safety constraint satisfaction, necessary to ensure the safe execution of the maneuvers. This paper introduces a control that addresses this limitation by integrating trajectory tracking, disturbance estimation, and safety enforcement. An extended high-gain observer is employed to estimate disturbances resulting from gravitational model uncertainties. We then apply a feedback-linearizing and disturbance-canceling controller that achieves exponential tracking of reference trajectories. Finally, we use a control barrier function based minimum-intervention controller to enforce state and input constraints through out the maneuver execution. This control combines trajectory tracking of offline generated reference trajectories with formal guarantees of safety, which follows common guidance and control architectures for spacecraft and allows aggressive maneuvers to be executed without compromising safety. Numerical simulations using fuel-optimal trajectories demonstrate the effectiveness of the controller in achieving precise and safe soft-landing, highlighting its potential for autonomous SCB missions.

Keywords: Aerospace, Constrained control, Optimal control
Abstract: This paper describes the application of a Shrinking Horizon Model Predictive Controller (SHMPC) to a cooperative spacecraft capture. SHMPC is used to coordinate multiple capturing spacecraft to specified non-equilibrium states corresponding to the capture of a rotating non-cooperative target subject to state and control constraints. Robustness to model mismatch, thrust uncertainty, and inexact optimization is illustrated through numerical simulations demonstrating the benefits of SHMPC over the following open-loop trajectories using conventional feedback. Enhancements to the baseline SHMPC formulation are proposed to reduce the computational effort.

Keywords: Predictive control for linear systems, Emerging control applications, Autonomous vehicles
Abstract: This letter presents a perception-driven safety filter that converts each 3D Gaussian Splat (3DGS) into a closed-form forward collision cone. The resulting constraint admits a relative-degree-one control barrier function (CBF) that is enforced online via a quadratic program (QP). By leveraging the analytic geometry of Gaussian splats, the proposed formulation yields a continuous, closed-form representation of collision avoidance constraints that is both simple and computationally efficient. In contrast to distance-based CBFs, which typically become active only when the robot is already near an obstacle, the collision cone CBF activates proactively, enabling earlier corrective action and thus smoother avoidance behavior at lower computational cost. The method is validated on a large synthetic scene consisting of approximately 170k Gaussian splats. Compared to a state-of-the-art 3DGS planner, the proposed filter reduces computation time by a factor of 3 and notably decreases trajectory jerk while maintaining safety. The approach is entirely analytic, requires no high-order CBF (HOCBF), and extends naturally to robots with physical extent via principled Minkowski-sum inflation of the splats. These properties make the method well suited for real-time navigation in cluttered, perception-derived environments on Earth and in space.

Keywords: Spacecraft control, Stability of linear systems, Control applications
Abstract: The deployment of large unfurlable reflectors presents significant challenges for spacecraft attitude stability, particularly when performed in open-loop (no control) mode due to uncertainties in deployment dynamics and limitations on active control authority. For the NASA–ISRO Synthetic Aperture Radar (NISAR) mission, a 12 m reflector was deployed in orbit using an open-loop sequence. To minimize drift from the Sun and ensure uninterrupted power generation, a momentumbiased configuration was implemented using reaction wheels. Unlike conventional spin-stabilization, which is unsuitable for reflector missions due to telemetry and telecommand pointing constraints, the reaction-wheel momentum bias provided controlled passive stability while maintaining spacecraft orientation. The bias value was derived through analytical modeling of spacecraft dynamics with reaction wheels under a smallangle approximation about the Earth-pointing reference frame, explicitly accounting for gravity-gradient torques and inertia properties. This analysis established stability margins and guided the selection of the commanded momentum bias. The approach also maximized the allowable open-loop duration, providing sufficient time for deployment even under contingency scenarios. Pre-flight predictions showed stable behavior with minimized solar and nadir drift, which was subsequently confirmed through on-orbit telemetry. Flight data demonstrated close agreement with analytical results, providing a flightproven validation of the momentum-biased attitude stability approach for large reflector deployment.

Keywords: Optimization algorithms, Computational methods, Aerospace
Abstract: A trajectory optimization method is devised for a class of problems with state constraints. The method is conceptually a projected gradient descent method, where the convergence of the control process is proved without reference to any discretization, and is obtained directly from the convergence of the cost function under mild regularity conditions. A corresponding algorithm for spacecraft trajectory optimization is proposed, along with two numerical examples where terminal states are prescribed. The algorithm is able to generate trajectories that require less fuel than stationary-action trajectories.

Keywords: Spacecraft control, Formal verification/synthesis, Optimization
Abstract: A key challenge in relative orbital motion is the design of nonlinear controllers that provide formal stability guarantees for the nonlinear dynamics while systematically incorporating operational constraints such as actuator saturation and sensor field-of-view requirements. In this paper, we present a computational framework for designing certified nonlinear controllers for orbital control under circular chief orbit. Our approach provides a unified solution leveraging three key steps: polynomial lifting to express the non-polynomial orbital dynamics in a polynomial form, the use of a dual Lyapunov theory for stability certification, and sum-of-squares optimization for tractable controller synthesis. Numerical simulations confirm effective constraint handling, including input saturation and geometric keep-in regions, and enhanced robustness to significant state measurement noise in comparison to LQR.

Keywords: Game theory, Agents-based systems, Emerging control applications
Abstract: Many shared public amenities such as transporta- tion networks and power grids suffer from congestion and depletion effects resulting from uncoordinated overconsumption by self-interested users. In response, researchers have long studied methods of encouraging pro-social consumption of shared societal resources. In this tutorial, we survey recent work in the area of behavior-influencing mechanisms in modern infrastructure-scale socio-technical systems. Two basic frame- works are considered. First, an informed social planner may use their informational advantage to influence behavior: our paper demonstrates a variety of situations in which interesting and counterintuitive principles emerge in this context. For instance, it is often the case that agents are better off if the planner reveals noisy information rather than acting with full transparency. We present a variety of distinct approaches and problem formulations for this information design setting, ranging from convergence results to analysis under nonstandard informational assumptions. Second, an infrastructure operator may leverage their monopoly power to use pricing policies to influence societal behavior. Here, we ask a robustness question and present recent work on methods of characterizing the sensitivity of social outcomes to variations in pricing policy.

Keywords: Network analysis and control
Abstract: Differential privacy has been used to privately calculate numerous network properties, but existing approaches often require the development of a new privacy mechanism for each property of interest. Therefore, we present a framework for generating entire networks in a differentially private way. Differential privacy is immune to post-processing, which allows for any network property to be computed and analyzed for a private output network, without weakening its protections. We consider undirected networks and develop a differential privacy mechanism that takes in a sensitive network and outputs a private network by randomizing its edge set. We prove that this mechanism does provide differential privacy to a network's edge set, though it induces a complex distribution over the space of output graphs. We then develop an equivalent privacy implementation using a modified Erdős–Rényi model that constructs an output graph edge by edge, and it is efficient and easily implementable, even on large complex networks. Experiments implement epsilon-differential privacy with epsilon of 2.5 when computing graph Laplacian spectra, and these results show the proposed mechanism incurs 49.34% less error than the current state of the art.

Keywords: Game theory, Network analysis and control, Optimization
Abstract: Self-optimizing behaviors can lead to outcomes where collective benefits are ultimately destroyed, a well-known phenomenon known as the ``tragedy of the commons". These scenarios are widely studied using game-theoretic approaches to analyze strategic agent decision-making. In this paper, we examine this phenomenon in a bi-level decision-making hierarchy, where low-level agents belong to multiple distinct populations, and high-level agents make decisions that impact the choices of the local populations they represent. We study strategic interactions in a context where the populations benefit from a common environmental resource that degrades with higher extractive efforts made by high-level agents. We characterize a unique symmetric Nash equilibrium in the high-level game, and investigate its consequences on the common resource. While the equilibrium resource level degrades as the number of populations grows large, there are instances where it does not become depleted. We identify such regions, as well as the regions where the resource does deplete.

Keywords: Adaptive control, Networked control systems, Distributed control
Abstract: We present a passivity-agnostic framework for distributed adaptive synchronization under position-only communication, bounded disturbances, and unknown leader dynamics. By passivity-agnostic we mean the design does not require the closed loop to be SPR a priori: it certifies SPR when present and recovers it by frequency shaping when absent. Followers are heterogeneous second-order systems with unknown (possibly unstable) dynamics. In the SPR regime, a structured reparameterization yields gradient-based adaptive error dynamics; Lyapunov analysis guarantees global asymptotic synchronization in the disturbance-free case, exact rejection of constant disturbances, and bounded responses to time-varying disturbances, with parameter convergence under persistent excitation. In the non-SPR regime, frequency shaping recovers effective passivity of the unshaped transfer function, enabling the same stability guarantees via standard passivity/Lyapunov arguments using Meyer-Kalman-Yakubovich (MKY) Lemma. Simulations across star, cyclic, path, and arbitrary graphs demonstrate scalable synchronization, robust tracking, and parameter adaptation under multiple disturbance profiles, confirming that the frequency-shaped non-SPR designs match the performance of the SPR case.

Keywords: Mean field games, Game theory, Control applications
Abstract: We study the control of rumor propagation in large networked populations by using Stackelberg graphon games. We first introduce a principal who wants to incentivize the spread of her preferred news and discourage the spread of non-preferred news. We define the Stackelberg graphon game equilibrium (SGGE), characterize the graphon game Nash equilibrium (GGNE) with a forward-backward differential equation system, and establish existence results. We further formulate a multi-leader model with two competing principals, each incentivizing her own preferred news. Finally, we propose a bi-level algorithm for computing (multi-leader) Stackelberg graphon game equilibria and conclude with numerical experiments where we show that existence competing principals will result in strong opinion divisions in the population.

Keywords: Network analysis and control, Markov processes, Estimation
Abstract: In network epidemic models, controlling the spread of a disease often requires targeted interventions such as vaccinating high-risk individuals based on network structure. However, typical approaches assume complete knowledge of the underlying contact network, which is often unavailable. While network structure can be learned from observed epidemic dynamics, existing methods require long observation windows that may delay critical interventions.

In this work, we show that full network reconstruction may not be necessary: control-relevant features, such as high-degree vertices (super-spreaders), can be learned far more efficiently than the complete structure. Specifically, we develop an algorithm to identify such vertices from the dynamics of a Susceptible-Infected-Susceptible (SIS) process. We prove that in an n-vertex graph, vertices of degree at least n^alpha can be identified over an observation window of size Omega (1/alpha), for any alpha in (0,1). In contrast, existing methods for exact network reconstruction requires an observation window that grows linearly with n. Simulations demonstrate that our approach accurately identifies super-spreaders and enables effective epidemic control.

Keywords: Networked control systems, Neural networks, Stability of nonlinear systems
Abstract: Local bifurcation analysis plays a central role in understanding qualitative transitions in networked nonlinear dynamical systems, including dynamic neural network and opinion dynamics models. In this article we establish explicit bounds of validity for the classification of bifurcation diagrams in two classes of continuous-time networked dynamical systems, analogous in structure to the Hopfield and the Firing Rate dynamic neural network models. Our approach leverages recent advances in computing the bounds for the validity of Lyapunov–Schmidt reduction, a reduction method widely employed in nonlinear systems analysis. Using these bounds we rigorously characterize neighbourhoods around bifurcation points where predictions from reduced-order bifurcation equations remain reliable. We further demonstrate how these bounds can be applied to an illustrative family of nonlinear opinion dynamics on k−regular graphs, which emerges as a special case of the general framework. These results provide new analytical tools for quantifying the robustness of bifurcation phenomena in dynamics over networked systems and highlight the interplay between network structure and nonlinear dynamical behaviour.

Keywords: Automotive systems, Automotive control, Control applications
Abstract: This study proposes a trailer anti-lock braking system (ABS) for commercial travel trailers, addressing the practical concern of unknown trailer parameters and the absence of trailer sensors. In current commercial trailer systems, a control unit in the towing vehicle allows the driver to manually adjust the trailer brake gain. However, improper adjustment can cause excessive braking at the trailer wheels, leading to wheel lock-up and a loss of lateral stability, commonly referred to as the “trailer swing” phenomenon. Previous studies are based on trailer wheel speed or hitch angle sensors, which are not typically available in commercial trailers, thus limiting their applicability. Aftermarket trailer ABS also exists, but its high cost and complex installation requirements have hindered widespread adoption. To overcome this limitation, this study introduces a novel trailer ABS that does not rely on direct wheel slip measurements. Instead, trailer swing is detected by the trailer swing index, an indirect indicator of wheel lock-up, enabling braking control without additional trailer-mounted sensors. The proposed method is evaluated through Carsim/Simulink. Through the proposed method, trailer swing is detected and controlled effectively without trailer information.

Keywords: Automotive systems, Mechatronics, Estimation
Abstract: With an increasing demand for Level 3 autonomous driving systems, the need for accurate tracking of the trajectories of nearby vehicles using just low-cost radar has become critical. While well-designed estimation algorithms for target vehicle tracking are essential, the accuracy of the underlying dynamic model plays a pivotal role as well. Conventional vehicle kinematic models are formulated in the inertial (global) frame and do not explicitly account for the ego-vehicle’s motion. The new relative motion model in this paper considers the effects of ego-vehicle motion, leading to improved target tracking performance when the ego-vehicle executes non-zero yaw motions. This model requires the use of the ego-vehicle’s dynamic states, including velocity, yaw rate, and slip angle. Among these, slip angle estimation poses a particular challenge due to limitations in cost-effective sensing technologies. This paper highlights the importance of slip angle estimation for accurate vehicle tracking using a new relative motion model. The ego-vehicle’s slip angles are employed for target vehicle tracking. The proposed method is validated across a range of driving scenarios, including sharp cornering maneuvers and even U-turns. Experimental results demonstrate that the proposed method consistently provides reliable vehicle tracking results.

Keywords: Automotive systems, Multivehicle systems, Estimation
Abstract: This paper introduces an observer design framework for a system with multivariable vector arguments in the nonlinear process dynamics. When the partial derivatives of the nonlinear functions are bounded and monotonic, exponential convergence of the observer is guaranteed by design using a set of linear matrix inequalities (LMIs). For non-monotonic cases, a switched-gain observer is proposed, complemented by a state-scaling technique that enhances numerical stability and extends the feasibility of the LMI conditions. The framework is validated in an autonomous driving application, where the observer estimates the trajectories of surrounding vehicles. Experiments on a full-scale autonomous Chrysler Pacifica vehicle at the University of Minnesota demonstrate robust performance even in the presence of non-monotonic vector nonlinear dynamics.

Keywords: Cooperative control, Automotive systems, Multivehicle systems
Abstract: Traditional platooning algorithms, such as Cooperative Adaptive Cruise Control (CACC), rely on connected automated vehicles (CAVs) sharing their current position, speed, and acceleration. In contrast, intent-based platooning is an emerging technology where CAVs share their planned trajectories. In this work, both of these algorithms are adapted for Battery Electric Vehicles (BEVs) and Plug-in Hybrid Electric Vehicles (PHEVs) to assess the impact of platooning on energy consumption. The regenerative braking system common in BEVs and PHEVs is modeled and incorporated into the optimization framework of intent-based platooning. The modified framework is then compared with the traditional CACC and intent-based platooning frameworks through simulations based on real vehicle speed profiles collected in free-flow and congested highway driving. The results show that the intent-based platooning framework adapted for BEVs and PHEVs achieves an 7-16% improvement in energy efficiency compared to the traditional intent-based platooning framework, and an 14-33% improvement compared to CACC.

Keywords: Optimization algorithms, Simulation, Automotive systems
Abstract: This work integrated a hybrid Bayesian–Optimisation and Genetic–Algorithm (BO–GA) workflow in the IPG CarMaker’s software-in-the-loop (SIL) environment to assist Advanced Driver Assistance System (ADAS) development. By replacing exhaustive grid searches with a Gaussian-process surrogate model and an expected-improvement acquisition function, the approach delivers substantial cost savings and delineates operating boundaries for subsequent software development. An Autonomous Emergency Braking (AEB) use case shows how the approach concentrates evaluations in the most critical regions of the scenario space, specifically where the timeto-collision (TTC) satisfies < 1.0s or the jerk exceeds 10m/s3 —thereby meeting both safety and comfort requirements. Relative to conventional parameter sweeps, the method cuts the total efforts by approximately 75% while still producing interpretable safety–comfort trade-off curves that guide further development.

Keywords: Multivehicle systems, Automotive control, Control applications
Abstract: Connected and automated vehicle (CAV) platooning promises gains in energy efficiency and traffic throughput and, most critically, in safety. These safety benefits hinge on string stability, which determines how disturbances propagate along a platoon. While longitudinal string stability is well studied, lateral string stability, which governs the propagation of path-tracking errors that can lead to unsafe deviations from the intended path, remains underexplored. Its importance is increasing as autonomous vehicles rely more heavily on onboard sensing and map‑free navigation, where sensor occlusion and dense formations amplify safety risks. This paper presents a new framework for lateral string stability that directly addresses safety‑critical path‑relative tracking errors and enables consistent comparison across vehicles following the same road geometry. Central to this framework is an arc‑length (Eulerian) viewpoint, a departure from traditional analyses, that clarifies how tracking errors at a given point on the path propagate from one vehicle to the next. A formal definition of lateral string stability is introduced along with two control strategies: an onboard‑sensing‑only controller and a novel learn‑from‑predecessor approach utilizing vehicle-to-vehicle (V2V) communication. We show that onboard sensing alone cannot guarantee attenuation of path-tracking errors, imposing a fundamental safety limitation, whereas V2V communication enables true error attenuation.

Keywords: Cooperative control, Agents-based systems, Optimization
Abstract: Aligning large language models (LLMs) with human preferences has been a critical challenge in artificial intelligence research. While Reinforcement Learning from Human Feedback (RLHF) is a popular approach to this alignment, it often relies on explicit human-provided rewards, which are costly, inconsistent, and difficult to scale. In this paper, we propose a comfort-guided supervised fine-tuning framework that replaces direct human feedback with reward signals computed from pre-built human comfort models. These models capture spatial, cognitive, and interactional preferences, enabling LLMs to generate decisions that align with human comfort expectations in collaborative scenarios. By fine-tuning LLMs using comfort-based and efficiency-based reward functions derived from these models, our method provides a scalable and systematic alternative to traditional feedback-driven learning. We apply this framework to a human-robot collaboration task, where an LLM generates optimal action plans by balancing efficiency and comfort. Our results demonstrate that fine-tuning with an extremely small dataset (1MB) and model-generated reward signals significantly improves alignment with human expectations, yielding decisions that are both effective and intuitively preferred by human collaborators. This study introduces a lightweight and efficient strategy to embed human comfort into AI-driven decision-making, reducing reliance on large-scale human annotation while maintaining high-quality preference alignment.

Keywords: Cooperative control, Agents-based systems, Optimization
Abstract: This paper presents an energy-based, pairwise collaborative scheduling framework for two robots with different mobility characteristics traversing a landscape. The collaborative scheduling problem is formulated as a mixed integer linear program (MILP), whose solution determines when the two robots should work together to accomplish their respective tasks with energy remaining in their batteries. The MILP is framed in two ways, considering net energy expenditure and final stored energy (i.e., energy available at the end of the task) as performance measures: (I) optimizing the performance of the overall team (e.g., minimizing the combined net energy expenditure of the two robots) and (II) optimizing the performance of individuals (e.g., minimizing the maximum net energy expenditure for the individual robots). Additionally, we introduce the notion of the ``cost of fairness'' and demonstrate the proposed framework on scenarios in which collaboration between two heterogeneous robots is necessary.

Keywords: Distributed control, Cooperative control, Decentralized control
Abstract: To address the challenge of efficient coverage by multi-robot systems in non-convex regions with multiple obstacles, this paper proposes a coverage control method based on the Generalized Voronoi Graph (GVG). This method partitions the non-convex region into multiple sub-regions using GVG, overcoming the limitations of traditional Voronoi partitioning, which fails to handle obstacles and non-convex boundaries. Besides, a weighted load-balancing algorithm is developed, which considers the quality differences among sub-regions. By iteratively optimizing the robot allocation ratio, the number of robots in each sub-region is matched with the sub-region quality to achieve load balance. The convergence of the method is proved and its performance is evaluated through simulations.

Keywords: Cooperative control, Distributed control, Control of networks
Abstract: This work introduces a distributed formation control strategy for multi-agent systems based solely on rotation symmetry constraints. We propose a potential function that enforces inter-agent rotational symmetries, whose gradient defines a control law that drives the agents toward a desired planar symmetric configuration. We show that only n-1 edges (the minimal connectivity requirement) are sufficient to implement the strategy, where n is the number of agents. We further augment the design to address the maneuvering problem, enabling the formation to undergo coordinated translations, rotations, and scaling along a predefined virtual trajectory. Simulation examples are provided to validate the effectiveness of the proposed method.

Keywords: Cooperative control, Reinforcement learning, Robust control
Abstract: The aim of this study is to propose a reinforcement-learning-based synthesis method for the integrated design of a robust H-infinity control and a deep neural network. This integration improves closed-loop performance through Proximal Policy Optimization (PPO), i.e. the Hinf controller and the deep neural network are trained within the same optimization process; this joint training constitutes the main novelty of the paper. The resulting control system provides performance guarantees against uncertainties which are taken into account both in the Hinf design process and in the training environment. The effectiveness of the integrated design method is illustrated through the coordinated control of simultaneously moving robot vehicles, where reduced mission time and collision avoidance are achieved even under significant uncertainties.

Keywords: Cooperative control, Autonomous robots, Sensor networks
Abstract: Autonomous robots are increasingly deployed to estimate spatiotemporal fields (e.g., wind, temperature, gas concentration) that vary across space and time. We consider environments divided into non-overlapping regions with distinct spatial and temporal dynamics, termed non-uniform spatiotemporal environments. Gaussian Processes (GPs) are commonly used to estimate such fields, where the kernel encodes spatial and temporal correlations through its lengthscales. When these lengthscales are misspecified, the resulting uncertainty estimates can be highly inaccurate. Existing GP methods often assume a single global lengthscale or update only periodically; some allow spatial variation but ignore temporal changes. To address these limitations, we propose a two-phase framework for multi-robot field estimation: Phase 1 uses a variogram-driven planner to learn region-specific spatial lengthscales, while Phase 2 reallocates robots based on current uncertainty and updates sampling as temporal lengthscales are refined. To encode uncertainty, we use clarity, an information metric from our earlier work. We evaluate the proposed method across diverse environments and provide convergence analysis for spatial lengthscale estimation, along with dynamic regret bounds quantifying the gap to the oracle allocation sequence.

Keywords: Estimation, Energy systems, Observers for nonlinear systems
Abstract: Lithium-ion batteries in parallel connection present significant challenges for monitoring and control due to the coupled dynamics of individual cells and the lack of direct cell-level sensing. The inability to monitor cell-level state-of-charge (SOC), individual branch currents, and key parameters limits the effectiveness of battery management strategies and compromises system safety. This paper proposes an adaptive observer framework that addresses these limitations through a descriptor system formulation capturing the coupled dynamics of parallel-connected cells while enforcing Kirchhoff's laws as algebraic constraints. The method enables simultaneous estimation of cell-level states and key parameters, particularly charge capacity, using exclusively total voltage and current measurements. Numerical simulations under a dynamic current profile demonstrate accurate reconstruction of SOC, cell-level currents, and charge capacities of individual cells, even with large initial condition mismatches and parameter heterogeneity. These results establish the potential of this observer-based solution for battery management systems seeking comprehensive cell-level awareness without additional sensing infrastructure.

Keywords: Machine learning, Aerospace, Energy systems
Abstract: Lithium plating during fast charging is a critical degradation mechanism that accelerates capacity fade and can trigger catastrophic safety failures. Recent work has shown that plating onset can manifest in incremental-capacity analysis as an additional high-voltage feature above 4.0 V, often appearing as a secondary peak or shoulder distinct from the main intercalation peak complex; however, conventional methods for computing dQ/dV rely on finite differencing with filtering, which amplifies sensor noise and introduces bias in feature location. In this paper, we propose a Gaussian Process (GP) framework for lithium plating detection by directly modeling the charge-voltage relationship Q(V) as a stochastic process with calibrated uncertainty. Leveraging the property that derivatives of GPs remain GPs, we infer dQ/dV analytically and probabilistically from the posterior, enabling robust detection without ad hoc smoothing. The framework provides three key benefits: (i) noise-aware inference with hyperparameters learned from data, (ii) closed-form derivatives with credible intervals for uncertainty quantification, and (iii) scalability to online variants suitable for embedded BMS. Experimental validation on Li-ion coin cells across a range of C-rates (0.2C-1C) and temperatures (0-40^circC) demonstrates that the GP-based method reliably resolves distinct high-voltage secondary peak features under low-temperature, high-rate charging, while correctly reporting no features in non-plating cases. The concurrence of GP-identified differential features, reduced charge throughput, capacity fade measured via reference performance tests, and post-mortem microscopy confirmation supports the interpretation of these signatures as plating-related, establishing a practical pathway for real-time lithium plating detection.

Keywords: Energy systems, Estimation, Machine learning
Abstract: Second-life batteries are critical to sustainable energy transi-tions, but their diverse historical usage history and limited field data pose major challenges for their reliable health as-sessment. To address this problem, we propose a pulse-based continual learning (PCL) framework with replay-buffer strategy that preserves prior knowledge while adapting to heterogeneous and data-scarce test conditions. To ensure the rapidness of data curation of second-life batteries, rapid pulse current injection are performed at varying state of charge (SOC) levels, widths, amplitudes, and polarizations. The extracted features from those response voltage signals are utilized as the input for PCL model training. The proposed PCL method treats these pulse current test conditions as do-main-incremental tasks and applies a replay-buffer strategy to balance model adaptation and knowledge retention. We performed extensive physical experiments on a broad set of second-life cells, encompassing multiple chemistries (NMC, LMO, LFP), capacities from 2.1 Ah to 35 Ah, and diverse physical forms. The results show that the average SOH esti-mation accuracy across experimental conditions is over 90%, while the forgetting measure for historical tasks is near-zero. The PCL effectively learns new tasks while retaining estab-lished knowledge, delivering accurate and consistent SOH estimates with evolving second-life conditions. Therefore, the proposed PCL method is robust and scalable, providing a pathway toward efficient second-life battery estimation, screening, regrouping with implications for sustainable re-using and recycling.

Keywords: Aerospace, Flight control, Optimization algorithms
Abstract: Unmanned aerial vehicles frequently operate in outdoor environments subject to wind disturbances. In areas where these wind disturbances are large, for example where wind speeds approach or surpass the maximum vehicle speed, the most energy efficient route from a start to a goal is often not a straight path. Improvements could potentially be obtained by working with the wind, rather than against it. This paper proposes a grid-free method to quickly solve multi-goal path planning problems in strong wind fields. By propagating an approximation of the reachable set boundary forward in discrete time, the method provides a way to determine trajectories between locations in strong wind fields. It also provides an approximation of the region reachable by the vehicle at after a given time interval, meaning that goals outside of this region that must be reached within that interval can be eliminated from consideration. Additionally, a time-based cost for travel between two locations can be determined by the number of set expansions required to connect them. Under the assumption that time of travel and energy consumption are inherently linked, these properties are applied to solve an energy-optimizing version of the traveling salesman problem (TSP) where the solution is the energy-optimizing path between all feasible goals within a strong wind field, namely one whose localized maximum speeds are faster than the vehicle's flight speed.

Keywords: Nonlinear systems identification, Energy systems, Automotive systems
Abstract: In this paper, we propose a sparse-frequency EIS measurement and reconstruction framework that (i) acquires impedance data at a limited number of frequencies and (ii) reconstructs the full spectrum using a Chebyshev-based pseudospectral formulation. The frequencies are chosen according to Chebyshev–Lobatto nodes mapped onto a logarithmic frequency axis. The coefficients of the Chebyshev basis are efficiently obtained through a Type-I discrete cosine transform (DCT-I) with end-point half-weighting. Based on this reconstruction, the parameters of the corresponding ECM are then derived analytically, eliminating the need for iterative Nonlinear Least Squares (NLLS) fitting. Furthermore, we provide an analytic characterization of the reconstruction error and establish its dependence on both the frequency range and the number of sampled points. Finally, we experimentally validate the approach on an 18650 lithium-ion cell, showing that sampling at only six frequencies between 0.1 Hz and 1000 Hz is sufficient to reproduce the impedance spectra with high fidelity. In addition, fewer measurements show a 70% reduction in measurement time. Across the full frequency band, the synthesized ECM faithfully overlays the measured Nyquist curves, achieving sub-milliohm absolute error.

Keywords: Modeling, Grey-box modeling, Energy systems
Abstract: We construct a physics-informed machine learning (PIML) model for rechargeable thick-electrode lithium-metal battery (LMB) cells by combining an enhanced single-particle model (SPMe) with a feedforward neural network (FNN). The hybrid model incorporates an additional pole necessary for modeling variation in electrochemical variables across the thickness of the electrode and guarantees bounded-input bounded-output stability by design. We train the model against the predictions of a Newman model simulated with the PyBaMM package. The trained PIML model predicts the voltage of a cell with a 200µm electrode at 1.8C with 93% less error than a plain SPMe model while remaining tractable for embedded applications. We provide Python code to simulate the Newman model, simulate the SPMe, and train the PIML model.

Keywords: Power systems, Control applications, Reinforcement learning
Abstract: This work proposes an adaptive optimal control scheme for DC shipboard power systems (SPS) with multiple distributed generators (DGs), leveraging a multilayer neural network (MNN) to enhance the performance of a supplementary energy storage system (ESS) under pulsed power load (PPL) conditions. PPLs impose large, short-duration power demands that challenge system stability due to nonlinear dynamics and operational constraints. The proposed reinforcement learning (RL) framework with lifelong hybrid learning (LHL) ad- dresses these challenges by pursuing three objectives: rapid ESS charging, DC bus voltage regulation, and proportional load current sharing among DG converters. An actor–critic MNN approximates the value function and control policy, where the critic network is updated using a hybrid scheme that refines weights both at and within sampling instants for faster convergence. A weight consolidation mechanism further enables lifelong learning, mitigating catastrophic forgetting and reducing overall cost, while the actor adapts using control input errors. Experimental validation on the OPAL-RT OP4512 hardware-in-the-loop (HIL) platform demonstrates that the proposed controller effectively achieves the control objectives under dynamic load conditions.

Keywords: Power systems, Smart grid, Machine learning
Abstract: Frequency stability is fundamental to the secure operation of power systems. With growing uncertainty and volatility introduced by renewable generation, secondary frequency regulation - responsible for restoring the nominal system frequency - must now deliver enhanced performance not only in the steady state but also during transients. This paper presents a systematic framework to embed learning in the design of a primal–dual controller that provides provable (potentially exponential) stability and steady-state optimality, while simultaneously improving key transient metrics, including frequency nadir and control effort, in a data-driven manner. In particular, we employ the primal-dual dynamics of an optimization problem that encodes steady-state objectives to realize secondary frequency control with asymptotic stability guarantee. To augment transient performance of the controller via learning, a change of variables on control inputs, which will be deployed by neural networks, is proposed such that under mild conditions, stability and steady-state optimality are preserved. It further allows us to define a learning goal that accounts for the exponential convergence rate, frequency nadir and accumulated control effort, and use sample trajectories to enhance these metrics. Simulation results validate the theories and demonstrate superior transient performance of the learning-augmented primal-dual controller.

Keywords: Power systems, Stability of linear systems, Decentralized control
Abstract: Integration of distributed energy resources has created a need for autonomous, dynamic voltage regulation. Decentralized Volt-VAr Control (VVC) of grid-connected inverters presents a unique opportunity for voltage management but, if designed poorly, can lead to unstable behavior when in feedback with the grid. We model the grid-VVC closed-loop dynamics with a linearized power flow approach, leveraging historical data, which shows improvement over the commonly used LinDistFlow model. This model is used to design VVC slopes by minimizing steady-state voltage deviation from the nominal value, subject to a non-convex spectral radius stability constraint, which has not been previously implemented within this context. We compare this constraint to existing convex restrictions and demonstrate, through simulations on a realistic feeder, that using the spectral radius results in more effective voltage regulation.

Keywords: Power systems
Abstract: In this paper, we present a Spatio-Temporal Graph with Deep Embedded Clustering framework (STaG-DEC) that jointly detects, localizes, and classifies hybrid threats. It employs a Graph Convolutional Attention Network (GCAT) to extract spatial dependencies and fuses the resulting embeddings with temporally differenced measurements via concatenation and linear projection before the Transformer encoder, which suppresses slow operational drift, highlights abrupt attack-induced transients, and preserves long-range dependencies. To address class imbalance and overlapping patterns, standard crossentropy with a DEC-based joint loss is proposed that promotes compact and better-separated latent clusters and stabilizes minority-class decision boundaries, thereby enabling precise node-level localization and classification across varying attack distributions. Experiments on IEEE 14-bus system demonstrate our superior detection performance with notably improved replay detection and robust behavior under imbalanced conditions.

Keywords: Smart grid, Power systems, Optimization algorithms
Abstract: Designing optimal radial configurations for microgrids involves Mixed-Integer Non-Linear Programming (MINLP) formulations that are often computationally prohibitive for large-scale systems. To overcome this, we reformulate the distribution optimization problem—where generator locations and capacities are predetermined—into a set-valued optimization framework. This allows us to leverage the recently proposed FORWARD algorithm to efficiently explore the radial configuration space while strictly adhering to physical power flow constraints. The resulting feasible configurations serve as high-quality suboptimal solutions or as effective warm-start points for conventional MINLP solvers. The key innovation lies in abstracting the microgrid network structure to enable rapid topological search while maintaining coupling with physical power flow requirements. Our approach introduces several strategic modifications: a phase-expanded node representation for multi-phase AC systems, physics-informed weight functions incorporating impedance characteristics, and constraint integration strategies that embed power system limitations directly within the sampling process. Numerical evaluations on IEEE benchmarks and synthetic networks of up to 400 nodes demonstrate polynomial-time performance, whereas traditional MINLP solvers often fail beyond 70 buses. The algorithm achieves solutions in seconds with competitive quality, enabling real-time microgrid management and large-scale distribution~planning.

Keywords: Modeling, Distributed parameter systems, Process Control
Abstract: This is an application paper that provides a simulator and a control algorithm for a thermal collection line in a parabolic trough collector for solar generation of electricity. The simulation is designed to match the TCP-100 facility of the Plataforma Solar de Almeria, a full-scale CIEMAT research center, in Spain. Of interest to the research community would be the reveal of control issues and a solution that includes two nonlinear algorithms, explanation of a simulator that can serve as a benchmark for control algorithm testing, and comprehensive issues that need to be included in evaluating control algorithms. Of interest to the practitioner community would be the use of simple nonlinear control techniques embedded in a traditional cascade structure to solve nonlinear and deadtime issues.

The simulator includes nominal direct normal irradiance (DNI) that changes over the day; random variations in DNI, optical efficiency, inlet temperature, ambient losses, and flow friction factor; and error on all measurements (calibration and noise).

Keywords: Robotics, Biomedical, Reinforcement learning
Abstract: Human locomotion involves complex coordination of over-actuated muscle systems and joints, making simulation and control design highly challenging. While recent reinforcement and imitation learning methods can replicate human-like kinematics, they often fail to produce physiologically realistic force patterns, largely due to the limited availability and consideration of reference force plate or electromyography (EMG) data. This paper presents a hybrid imitation learning framework that integrates muscle-driven simulations with reinforcement learning to address over-actuation and to account for the fact that the available data reflects closed-loop actions involving muscle control. A key contribution is the incorporation of metabolic cost into the reward function, shaping energetically efficient and physiologically plausible controllers. Simulation results demonstrate that the learned policies generate muscle activations and ground reaction forces that align more closely with OpenSim references and experimental data than standard imitation learning. The method provides a scalable tool for developing and validating closed-loop control strategies for assistive systems such as exoskeletons and prostheses, while highlighting broader implications for learning-based control of over-actuated biomechanical systems.

Keywords: Robotics, Control applications, Machine learning
Abstract: Existing control designs for underactuated balance robots often rely on static equilibrium, restricting system autonomy. While dynamic equilibrium resolves this, it is computationally demanding to obtain, especially under unknown dynamics. This paper proposes a machine learning-based dynamic equilibrium prediction framework that captures inherent dynamic coupling directly from motion data, requiring no explicit equilibrium data for training. Leveraging this predicted equilibrium improves control design, enabling simultaneous task execution of the actuated subsystem and robust balance stabilization of the unactuated subsystem. Simulations validate the framework's effectiveness and enhanced performance.

Keywords: Robotics, Modeling, Predictive control for nonlinear systems
Abstract: This paper presents a novel multirotor design featuring telescoping extension arm morphology (TEAM), together with a generalized control allocation framework. TEAM enables continuous and asymmetric arm extension, offering a level of adaptability not achievable with prior morphing designs, which were constrained to symmetric or discrete arm configurations. To manage the nonlinear actuator interactions introduced by TEAM, we introduce a joint mapping loop based on model predictive control (MPC). This formulation simultaneously determines propeller speeds and arm lengths, providing optimal control allocation in real time. The framework integrates seamlessly with standard cascaded flight control architectures: conventional position and attitude controllers remain unchanged, while MPC operates as a third loop with minimal structural modification. This combined design–control approach represents the first generalized framework for morphing multirotors with fully telescoping arms. The simulation results demonstrate that the proposed system achieves accurate trajectory tracking under aggressive maneuvers and center-of-mass offsets caused by payloads, highlighting its potential for robust and efficient aerial logistics.

Keywords: Robotics, Human-in-the-loop control, Simulation
Abstract: Dexterous in-hand manipulation (DIM) remains a major challenge for prosthetic hand control, as users must generate complex multi-joint motions from low-dimensional inputs. Standard principal component analysis (PCA) captures static postural synergies but ignores continuous movement over time, limiting smooth finger coordination and practical prosthetic control. We apply multivariate functional PCA (mfPCA) to model joint trajectories as multivariate functional data, capturing both the temporal dynamics and the inter-joint coordination. Simulations with a high-degree-of-freedom robotic hand show that, with sufficient components, mfPCA-reconstructed trajectories closely reproduce the original object manipulation at joint and task levels. To enable practical low-dimensional control, we introduce a clustering-based strategy that maps synergy weights onto a low-dimensional plane, supporting stable task reproduction and adjustable variation. These results provide a framework for natural, low-burden prosthetic hand control in DIM tasks.

Keywords: Robotics, Predictive control for nonlinear systems, Control applications
Abstract: Control input for the actuated subsystem and unactuated subsystem in underactuated robots’ control is separated into two groups. A coordinated scheme is proposed to design control for each subsystem to achieve trajectory tracking and stabilization, including a computed torque control for actuated subsystem tracking and an MPC-based control for unactuated subsystem balance. The MPC-based scheme in a subspace simultaneously predicts equilibrium and solves control design by embedding the unactuated dynamics as a nonlinear algebraic constraint. The effectiveness of the coordinated control is verified by extensive validations and comparisons.

Keywords: Robotics, Control applications, Optimal control
Abstract: Humanoid robots must maintain stability under external forces during loco-manipulation tasks such as cart pushing or ladder carrying. This paper proposes a model predictive control (MPC) framework that extends the Linear Inverted Pendulum Model (LIPM) by incorporating external wrenches as additive offsets in the zero-moment point (ZMP) formulation. Through an equivalent ZMP transformation, the approach preserves the linear and convex structure of the MPC problem, enabling real-time feasibility while retaining physical interpretability. The method was validated in MuJoCo simulations with a full-size humanoid robot under various manipulation forces. Results demonstrate stable locomotion during lateral disturbances, ladder transportation, wheelbarrow maneuvering, and heavy-object pushing. These findings show that the proposed framework bridges the gap between efficient LIPM-based walking and robust multi-contact manipulation, advancing humanoid capability in real-world tasks.

Keywords: Discrete event systems, Automata, Supervisory control
Abstract: This paper introduces a mathematical framework that helps us understand when a discrete event system (DES) begins to experience blocking due to transition deletions. We have developed a string-based analysis to determine if non-blocking conditions persist following these deletions. Our contributions can be summarized in three key points: a state-specific, necessary-and-sufficient characterization of blocking caused by transition deletions, a definition of robust deletions that maintain non-blocking conditions, and an algorithm that identifies critical transitions through minimal blocking sets. This algorithm streamlines computation by focusing on structured candidates instead of considering all possible deletions. We present case studies in manufacturing and autonomous vehicles that effectively identify critical transitions and blocking scenarios while significantly reducing complexity.

Keywords: Discrete event systems, Optimization, Automata
Abstract: This work addresses the problem of protecting multiple global secrets in discrete event systems modeled by nondeterministic finite automata. Specifically, a global secret in a system is assumed to be composed of one or more states, each assigned a specific security level. We say that a state is protected if any sequence of events leading to it from the initial state contains a quantity of protected events that is equal to or greater than the required security level. Correspondingly, a global secret is said to be protected if the cumulative weight of its containing protected states (or all containing states are considered in the worst-case scenario) satisfies a user-defined protection threshold. Our objective is to develop an event protection policy that is capable of protecting global secrets. To do so, we build an augmented automaton which eliminates the difficulty of analyzing the infinite sequences when verifying protection. Then, an optimal protection policy with minimum cost can be obtained by solving an integer linear programming problem. Examples are given to illustrate the effectiveness of our proposed protection strategy.

Keywords: Discrete event systems, Petri nets, Process Control
Abstract: This paper proposes a set of novel sufficient conditions that guarantee three behavioral properties in signal interpreted Petri nets (SIPNs), i.e., liveness, reachability, and reversibility. SIPNs provide a modeling formalism for representing the control algorithm of discrete manufacturing systems. The liveness, reachability, and reversibility properties ensure that the desired control actions remain perpetually executable, the system states are reachable, and that the system can always return to its initial state, respectively. The sufficient conditions are derived based on the Petri net state equation and the enabling rules of the transitions in SIPNs. Moreover, it is shown how the reachability and reversibility can be computationally verified based on an integer linear programming problem. The computational complexity for verifying the properties is polynomial with respect to the number of markings in the SIPN when the reachable set is already available.

Keywords: Discrete event systems, Supervisory control, Automata
Abstract: Large-scale systems found in technological frameworks often consist of interacting modules, where ensuring security is critical. In discrete-event systems, normality ensures that, based on an system observations, one can unambiguously determine whether the corresponding plant behavior is safe. In this paper, we first investigate the computation of an earlier form of normality, known as (L,P)-normality, in modular discrete-event systems. We show that, under the condition that all shared events are observable, the supremal (L,P)-normal sublanguages can be computed in a modular fashion. We then apply these results to the enforcement of two related notions that share similar physical interpretations with (L,P)-normality---namely, critical observability and weak opacity. By integrating our contributions with existing results on controllability and normality, we develop a modular supervisory control approach to enforce these properties and provide examples to illustrate the enforcement framework.

Keywords: Networked control systems, Stability of nonlinear systems, Delay systems
Abstract: Nonlinear positive system networks with heterogeneous time-delay communications exhibit strong nonlinear effects that are highly sensitive to delays, which complicates their analysis and limits the generalisation of existing results. This paper investigates the stability and well-posedness of such networks. We first establish two fundamental properties of the considered dynamics. Well-posedness guarantees that communication delays do not introduce singularities into the solution and ensures that, for any admissible initial conditions, the system admits a unique trajectory evolving continuously over time. Positivity ensures that all trajectories remain non-negative for admissible initial conditions. Building upon these properties, we develop a new analysis framework that employs the theory of sequences of functions to construct a less-conservative linear Lyapunov function candidate. We show that stability hinges on the definiteness of a newly derived matrix-valued function guided by Barbalat's theorem. The theoretical results are validated through numerical simulations, demonstrating the effectiveness of the obtained results.

Keywords: Agents-based systems, Autonomous robots, Autonomous systems
Abstract: Recent machine learning research has increasingly focused on equilibrium analysis in non-cooperative games rather than solely on optimal solutions. Many such problems involve shared constraints and can be formulated as Generalized Nash Equilibrium Problems (GNEPs). For strongly monotone games, existing methods compute consensus-based variational GNEs (v-GNEs) by exchanging Lagrange multipliers. We propose a fully distributed continuous-time algorithm for shared linear equality constraints that converges without multiplier exchange and reaches any GNE, reducing communication overhead and improving privacy. Discrete-time schemes are also provided, and the method is validated on a multi-robot placement task.

Keywords: Agents-based systems, Autonomous systems, Cooperative control
Abstract: Multi-agent UAV systems are well-suited for large-scale data collection and transportation, though navigation in unstructured, GPS-denied environments remains challenging. Vision-based navigation enables operation without external infrastructure, but the uncertainty in state estimation limits its reliability in multi-agent settings. We propose a trajectory planning framework that incorporates estimator uncertainty by exploiting visual feature observations between agents. The framework maintains a coherent shared map through multi-agent frame alignment to prevent independent vision drift, and employs a perception-aware reward that favors trajectories with stronger feature visibility and cross-agent redundancy. Flight data from a controlled two-UAV experiment demonstrate that our alignment module can effectively reduce relative distance error, validating its role in maintaining inter-agent consistency. Simulations show that perception-aware rewards improve feature visibility and coordination while maintaining goal-reaching performance.

Keywords: Agents-based systems, Autonomous systems, Discrete event systems
Abstract: In multiplayer games with sequential decision-making, self-interested agents may form and dissolve temporary coalitions to maximize satisfaction of temporal goals beyond individual capabilities. We study a class of multiplayer games played on graphs where, in each round, a designated leader declares a coalition it joins and prescribes a joint action. Coalition members must comply unless their assigned action is inadmissible. The remaining agents act independently using admissible strategies. Each agent has temporal goals prioritized by qualitative preferences. We present a unified framework that integrates coalition formation with rational strategy synthesis under partial-order preferences. First, we generalize classical admissibility by introducing maximal sure-winning strategies to track goal satisfaction across evolving coalitions. Second, we define a multi-dimensional value function that captures preference-aware guarantees for all agents and enables efficient coalition rationality analysis. Third, we develop a polynomial-time algorithm to synthesize admissible strategies for all agents and proves their existence for this class of games. Experiments in a blocks-world domain demonstrate the advantages of dynamic over fixed coalitions, revealing counter-intuitive behaviors: agents with aligned preferences may fail to cooperate, while those with conflicting goals may form beneficial coalitions.

Keywords: Agents-based systems, LMIs, Distributed control
Abstract: This paper investigates the distributed output feedback control problem for heterogeneous multi-agent systems under a leader-following architecture with guaranteed L_2 performance. A scalable synthesis framework is developed, where the design of the leader state observer and the local dynamic output feedback controller are decoupled, thereby significantly reducing the computational complexity for large-scale networks. The proposed approach enables each agent to implement the controller using only local information.

Linear matrix inequality conditions are derived to guarantee desirable tracking and disturbance attenuation. The dimension of the LMIs to be solved does not increase with the number of agents. The feasibility condition is established, while the effectiveness is demonstrated through a simulation example.

Keywords: Agents-based systems, Cooperative control, Distributed control
Abstract: In this paper, we propose a distributed model predictive control (DMPC) framework for cooperative formation of multi-agent systems that composed of unmanned aerial vehicles (UAVs) and an unmanned ground vehicle (UGV). The designed system considers formation maintenance, collision avoidance under information delays, uncertain deployments, and obstacle environments. A linear recursive estimation with error-shift compensation reconstructs the delayed neighbor trajectories, and the risk-aware re-optimization mechanism refines control when inter-agent distances fall below a threshold. Simulations results along with real-world experiments are provided to demonstrate robust formation tracking and collision avoidance.

Keywords: Agents-based systems, Reinforcement learning, Optimal control
Abstract: The paper designs data-driven distributed optimal consensus control for networked multiple dynamic systems, with each system (representing an agent) having multiple control inputs (representing players). We formulate differential graphical games that incorporate the evaluation of continuous-time multiplayer multiagent systems (MMSs), the cost functions of all players in all agents, and the communication network. The min-max strategy is designed for each player to prepare the optimal response to the worst-case reactions from the other players in the same and different agents. This strategy provides each player with distributed algebraic Riccati equations (AREs) and designs the distributed optimal control based on local feedback, further certified by ensured solutions in AREs, the achievement of min-max strategy, and the asymptotic stability of local error systems. The model-free, data-driven reinforcement learning (RL) algorithm is further developed to solve the min-max solutions using online sampled local trajectories without knowing system dynamics. Simulations verify the strategy and algorithm.

Keywords: Statistical learning, Decentralized control, Agents-based systems
Abstract: We develop Dif-AltGDmin, a diffusion-based decentralized federated algorithm for multi-task representation learning (MTRL). We consider multi-task linear regression where multiple models share a common low-dimensional representation, and propose an alternating projected gradient descent and minimization approach to recover the shared low-rank feature matrix without a central coordinator. We provide provable guarantees on sample and iteration complexity to achieve ε-accuracy, and show that Dif-AltGDmin achieves significantly reduced communication complexity compared to existing decentralized approaches. Numerical simulations confirm that our algorithm matches the convergence of its centralized counterpart while remaining communication-efficient.

Keywords: Statistical learning, Optimization, Stochastic optimal control
Abstract: Computing the similarity between two probability distributions is a recurring theme across control. We introduce a unified family of distances between the probability distributions of two random variables that is based on the discrepancy between the cumulative distribution functions of random linear one-dimensional projections of the random variables. Our proposed distance is interpretable, computationally simple, and admits a differentiable approximation. We establish asymptotic theoretical guarantees for sample-based estimators of the distance. We empirically study the use of the distance in a two-sample test and demonstrate its ability to distinguish different distributions. Finally, we show that the distance allows for simple gradient-based solutions in control by studying distribution steering and ergodic control.

Keywords: Statistical learning, Stochastic systems, Kalman filtering
Abstract: This paper studies the problem of online multi-step-ahead prediction for unknown linear stochastic systems. Using conditional distribution theory, we derive an optimal parameterization of the prediction policy as a linear function of future inputs, past inputs, and past outputs. Based on this characterization, we propose an online least-squares algorithm to learn the policy and analyze its regret relative to the optimal model-based predictor. We show that the online algorithm achieves logarithmic regret with respect to the optimal Kalman filter in the multi-step setting. Furthermore, with new proof techniques, we establish an almost-sure regret bound that does not rely on fixed failure probabilities for sufficiently large horizons N. Finally, our analysis also reveals that, while the regret remains logarithmic in N, its constant factor grows polynomially with the prediction horizon H, with the polynomial order set by the largest Jordan block of eigenvalue 1 in the system matrix.

Keywords: Reinforcement learning, Hierarchical control, Agents-based systems
Abstract: Multi-agent reinforcement learning (MARL) systems face fundamental challenges in balancing global coordination with local execution across different temporal scales. This paper introduces the Coupled Hierarchical Multi-Agent System (CHMAS), a novel framework that decomposes multi-agent decision-making into centralized strategic planning and distributed tactical execution with bidirectional information flow. The strategic layer integrates all agents' states with an exclusive global environmental state to generate guidance actions every T timesteps, while tactical agents execute distributed policies augmented by strategic guidance and local neighborhood observations. Unlike existing hierarchical approaches with unidirectional control, CHMAS establishes a feedback mechanism where accumulated tactical rewards influence strategic objectives through a coupling coefficient lambda, ensuring strategic plans remain grounded in tactical feasibility. To address the non-stationarity inherent in hierarchical learning, we propose an asynchronous update protocol where strategic parameters update every N_f tactical episodes, allowing tactical policies to converge to quasi-stationary points between strategic changes. We present both a general bi-level formulation capturing full system dynamics and a tractable additive approximation enabling rigorous analysis. Theoretical analysis proves that this asynchronous scheme achieves mathcal{O}(log K/sqrt{K}) convergence for the strategic layer after K strategic updates under standard assumptions. Experimental validation in a multi-agent foraging domain demonstrates successful learning of spatially partitioned exploration strategies, with both layers converging stably despite hierarchical coupling.

Keywords: Reinforcement learning, Learning, Machine learning
Abstract: Offline reinforcement learning (RL) enables policy optimization from fixed datasets, making it suitable for safety-critical applications where online exploration is infeasible. However, these datasets are often contaminated by adversarial poisoning, system errors, or low-quality samples, leading to degraded policy performance in standard behavioral cloning (BC) and offline RL methods. This paper introduces Density-Ratio Weighted Behavioral Cloning (Weighted BC), a robust imitation learning approach that uses a small, verified clean reference set to estimate trajectory-level density ratios via a binary discriminator. These ratios are clipped and used as weights in the BC objective to prioritize clean expert behavior while down-weighting or discarding corrupted data, without requiring knowledge of the contamination mechanism. We provide theoretical guarantees on convergence to the clean policy, finite-sample bounds, and robustness under varying contamination levels. A comprehensive evaluation framework is established, incorporating diverse poisoning protocols (reward, state, transition, and action) on D4RL continuous control benchmarks. Experiments demonstrate that Weighted BC maintains near-optimal performance even at high contamination ratios outperforming baselines such as traditional BC, BCQ, and BRAC.

Keywords: Neural networks, Game theory
Abstract: Bias pervades classical approximation methods in machine learning, including neural networks, kernel-based estimators, and density estimators, as well as advanced techniques such as multi-level Monte Carlo when applied to density-dependent functionals. Although some approaches can be rendered asymptotically unbiased under infinite computational budgets, such conditions are unattainable in practice. This limitation is particularly acute in mean-field-type games with Nemytskii-type state coefficients and payoffs, which arises in generative machine intelligence in audio, video, and image domains. Existing methodologies for these problems such as score function estimation inherit and propagate structural bias through all stages of computation. Here we introduce a debiasing framework for mean-field-type transformers that eliminates bias across these settings.

Keywords: Stability of nonlinear systems, Lyapunov methods
Abstract: This paper introduces a novel Lyapunov-based small-gain methodology for establishing fixed-time stability (FxTS) guarantees in interconnected dynamical systems. Specifically, we consider interconnections in which each subsystem admits an individual fixed-time input-to-state stability (ISS) Lyapunov function that certifies FxT-ISS. We then show that if a nonlinear small-gain condition is satisfied, then the entire interconnected system is FxTS. Our results are analogous to existing Lyapunov-based small-gain theorems developed for asymptotic and finite-time stability, thereby filling an important gap in the stability analysis of interconnected dynamical systems. The proposed theoretical tools are further illustrated through analytical and numerical examples, including the first result on fixed-time feedback optimization of dynamical systems without time-scale separation between the plant and the controller.

Keywords: Stability of nonlinear systems, Adaptive control, Optimization
Abstract: In this paper, we show uniform semi-global practical asymptotic stability of fast extremum seeking control (ESC) for single-input single-output Wiener systems. While classic ESC requires a time-scale separation between plant and dither, the fast ESC method circumvents this time-scale separation by exploiting limited knowledge of the frequency response of the linear part of the Wiener system, thereby achieving faster convergence. The assumptions under which the fast ESC method works are relaxed compared to existing work and explicit bounds on the design parameters of the fast ESC scheme are provided. A numerical case study illustrates the enhanced convergence and the robustness of the fast ESC method.

Keywords: Stability of nonlinear systems, Computational methods, Power electronics
Abstract: The classical asymptotic gain (AG) is a concept known from the input-to-state stability theory. Given a uniform input bound, AG estimates the asymptotic bound of the output. Sometimes, however, more information is known about the input than just a bound. In this paper we consider the case of a periodic input. Under the assumption that the system converges to a periodic solution, we introduce a new gain, called period-aware asymptotic gain (PAG), which employs periodicity to enable a sharper asymptotic estimation of the output. Since the PAG can distinguish between short-period ("high-frequency") and long-period ("low-frequency") signals, it is able to rigorously quantify such properties as bandwidth, resonant behavior, and high-frequency damping. We discuss how the PAG can be computed and illustrate it with a numerical example from the field of power electronics.

Keywords: Stability of nonlinear systems, Hybrid systems, Optimization
Abstract: We study the instability of Nesterov's ODE in non-conservative settings, where the driving term is not necessarily the gradient of a potential function. While convergence properties under Nesterov's ODE are well-characterized for settings with gradient-based driving terms, we show that the presence of emph{arbitrarily small} non-conservative terms can lead to instability. To resolve the instability issue, we study a regularization mechanism based on restarting. The resulting system is a hybrid dynamical system that mirrors Nesterov's ODE during intervals of flow, and implements restarts of the momentum state through discrete periodic jumps. For this system, we establish novel explicit bounds on the resetting period that ensure the decrease of a suitable Lyapunov function, guaranteeing stability as well as "accelerated" convergence rates under suitable smoothness and monotonicity properties on the driving term. Numerical simulations support our results.

Keywords: Stability of nonlinear systems, Lyapunov methods, Constrained control
Abstract: We show that the existence of a strictly compatible pair of control Lyapunov and control barrier functions is equivalent to the existence of a single smooth Lyapunov function that certifies both asymptotic stability and safety. This characterization complements existing literature on converse Lyapunov functions by establishing a partial differential equation (PDE) characterization with prescribed boundary conditions on the safe set, ensuring that the safe set is exactly certified by this Lyapunov function. The result also implies that if a safety and stability specification cannot be certified by a single Lyapunov function, then any pair of control Lyapunov and control barrier functions necessarily leads to a conflict and cannot be satisfied simultaneously in a robust sense.

Keywords: Nonlinear output feedback, Control applications, Stability of nonlinear systems
Abstract: This paper presents a novel algorithm for the dynamic inversion of nonlinear systems within the output tracking control framework, where the inverse serves as an open-loop nominal controller to facilitate feedback stabilization of the tracking error dynamics along a nominal trajectory. The desired dynamic response for the open-loop controlled system is treated as a pseudo identity whereby the causal and stable pseudo inverse is defined. For non-minimum phase systems, the Fliess canonical form is used to enable a practical pseudo inverse design without internal dynamics stabilization or reverse-time previewing. Unlike exact inversion, which does not have the desired dynamics, and can be unstable or anti-causal, the proposed pseudo inverse not only provides the desired dynamics of the open-loop controlled systems but is also stable and implementable in real time. A nontrivial application example illustrates the algorithm.

Keywords: Neural networks, Machine learning, Formal verification/synthesis
Abstract: Recurrent neural networks (RNNs) are widely employed to model complex dynamical systems due to their hidden-state structure, which inherently captures temporal dependencies. This work presents a hybrid zonotope–based approach for computing exact forward and backward reachable sets of closed-loop RNN systems with ReLU activation functions. The method formulates state-pair sets to compute reachable sets as hybrid zonotopes without requiring unrolling. To improve scalability, a tunable relaxation scheme is proposed that ranks unstable ReLU units across all layers using a triangle-area score and selectively applies convex relaxations within a fixed binary limit in the hybrid zonotopes. This scheme enables an explicit trade-off between computational complexity and approximation accuracy, with exact reachability as a special case.

Keywords: Neural networks, Constrained control, Data driven control
Abstract: In this letter, we present a neural network (NN)-based online adaptive controller for an n-link robot manipulator. The proposed controller utilizes a control barrier function (CBF) to enforce joint position and velocity constraints while simultaneously guaranteeing accurate trajectory tracking of the robot. In particular, we first map the constraints from the state space into the filtered-tracking error space. We demonstrate that maintaining forward invariance of the constrained error set guarantees constraint satisfaction in the state space. Additionally, we analytically establish that this transformation creates a boundary-to-boundary relationship between the two sets. The resulting NN-driven control torque ensures accurate trajectory tracking of the closed-loop robot manipulator while adhering to the prescribed state constraints even in the presence of uncertain dynamics. Finally, we validate the effectiveness of the proposed control scheme both analytically and numerically.

Keywords: Neural networks, Adaptive control, Robotics
Abstract: This paper introduces a novel safety-critical adaptive spiking neural network (SNN) control framework for uncertain nonlinear systems. On the safety side, we formulate the Augmented Barrier States (ABS) methodology into a unified safety-embedded tracking framework, establishing rigorous safety equivalence results that embed safety constraints directly within the closed-loop dynamics. On the learning side, a deep SNN architecture is developed with online adaptation enabled through direct error-driven weight update laws applied at every layer. Unlike conventional gradient approaches that struggle with instability due to the discontinuous nature of spike generation, the proposed method incorporates singular value decomposition (SVD) to regularize gradient flow, enhancing stability and convergence during online learning. The resulting controller guarantees uniformly ultimately bounded tracking while strictly preserving safety constraints for nonlinear systems, ensuring both adaptability and formal safety guarantees. Simulation studies on two-link manipulator validate the framework, demonstrating robust safety preservation, efficient real-time adaptation, and 80% computational energy savings compared to existing ANN-based approaches.

Keywords: Neural networks, Nonlinear systems identification, Power systems
Abstract: Modeling dynamical systems is crucial across the science and engineering fields for accurate prediction, con- trol, and decision-making. Recently, machine learning (ML) approaches, particularly neural ordinary differential equations (NODEs), have emerged as a powerful tool for data-driven modeling of continuous-time dynamics. Nevertheless, standard NODEs require a large number of data samples to remain consistent under varying control inputs, posing challenges to generate sufficient simulated data and ensure the safety of control design. To address this gap, we propose trajectory-sensitivity- aware (TRASE-)NODEs, which construct an augmented system for both state and sensitivity, enabling simultaneous learning of their dynamics. This formulation allows the adjoint method to update gradients in a memory-efficient manner and ensures that time-invariant control set-point effects are captured in the learned dynamics. We evaluate TRASE-NODEs using damped oscillator and inverter-based resources (IBRs). The results show that TRASE-NODEs generalize better from the limited training data, yielding lower prediction errors than standard NODEs for both examples. The proposed framework offers a data-efficient, control-oriented modeling approach suitable for dynamic systems that require accurate trajectory sensitivity prediction.

Keywords: Neural networks, Optimization algorithms, Emerging control applications
Abstract: Certification of neural networks supports their integration into commercial and safety-critical applications with formal guarantees. The Lipschitz constant, often used to certify neural networks, is a problematic proxy for robustness. Recent work on adversarial attacks has made it clear that the output of neural networks can change dramatically under minor perturbations. We expand the use of the hybrid zonotope representation of ReLU feedforward neural networks to calculate the minimum perturbation required to change a network's classification.

Keywords: Hybrid systems, Neural networks
Abstract: This paper establishes approximation results for neural network approximators of hybrid dynamical systems. We propose neural hybrid equations, which are hybrid systems that use neural networks as the flow map and the jump map while preserving the hybrid structure. Our main result proves that solutions to neural hybrid equations, with Lipschitz non-affine activation functions, can approximate solutions of nominal hybrid systems with Lipschitz vector fields with arbitrary precision on compact time domains. For the subclass of neural hybrid equations with rectifying linear unit (ReLU) activation functions, we establish O(1/N) convergence rates where N is the number of neurons. We illustrate our results with a numerical example.

Keywords: Stochastic optimal control, Emerging control applications, Autonomous systems
Abstract: Safety-critical control is imperative for deploying autonomous systems in the real world. Control Barrier Functions (CBFs) offer strong safety guarantees when accurate system and sensor models are available. However, widely used additive, fixed-noise models are not representative of complex sensor modalities with state-dependent error characteristics. Although CBFs have been designed to mitigate uncertainty using fixed worst-case bounds on measurement noise, this approach can lead to overly-conservative control. To solve this problem, we extend the Belief Control Barrier Function (BCBF) framework to accommodate state-dependent measurement noise via the Generalized Extended Kalman Filter (GEKF) algorithm, which models measurement noise as a linear function of the state. Using the original BCBF framework as baseline, we demonstrate the performance of the BCBF-GEKF approach through simulation results on a 1D single integrator setpoint tracking scenario and a trajectory tracking scenario using 2D unicycle kinematics. Our results confirm that the BCBF-GEKF approach offers less conservative control with greater safety.

Keywords: Stochastic optimal control, Switched systems, Optimization
Abstract: In this paper, we solve the chance-constrained covariance steering problem for discrete-time Markov Jump Linear Systems (MJLS) using a convex optimization framework. We derive the analytical expressions for the mean and covariance trajectories of time-varying discrete-time MJLS and show that they cannot be separated even without chance constraints, unlike the single-mode (non-hybrid) dynamics case. To solve the covariance steering problem, we propose a two-step convex optimization framework and then incorporate chance constraints through an iterative framework. Since the stochastic hybrid nature of MJLS introduces non-Gaussian distributions unlike single-mode covariance steering, we also propose deterministic formulations for non-Gaussian chance constraints based on Cantelli's inequality and the multivariate Chebyshev inequality. Both problems are originally nonconvex, and we derive convex relaxations which are proved to be lossless at optimality using the Karush–Kuhn–Tucker (KKT) conditions. Numerical simulations demonstrate the proposed method by achieving target covariances while respecting chance constraints under additive noise, bias, and Markovian jump dynamics.

Keywords: Stochastic optimal control, Optimal control, Stochastic systems
Abstract: This paper examines a two-player, zero-sum stochastic differential game problem over an infinite time horizon, where the players employ controller (pursuer) and stopper (evader) policies for a nonlinear system driven by Brownian motion. The pursuer's goal is to minimize a nonlinear-nonquadratic performance criterion while guaranteeing stochastic stabilization and the evader's goal is to maximize the performance criterion. By establishing a connection between stochastic Lyapunov stability theory and the stochastic Hamilton-Jacobi-Isaacs equation, we derive explicit optimal strategies for both players that guarantee stochastic stabilization as well as enforcing a saddle point condition on a nonlinear nonquadratic performance criterion. We demonstrate that global asymptotic stability in probability is ensured through a Lyapunov function, which also serves as the solution to the steady-state stochastic Hamilton-Jacobi-Isaacs equation, thereby guaranteeing both closed-loop stability in probability and optimality. Furthermore, we develop optimal feedback controller and stopper policies for affine nonlinear systems using an inverse optimality framework tailored to the game problem. These results extend existing linear feedback pursuer-evader policies to nonlinear controllers and stoppers that optimize general polynomial and multilinear performance criteria.

Keywords: Stochastic optimal control, Optimal control, Stochastic systems
Abstract: This paper presents a unified framework for the optimal nonlinear analysis and feedback control design of stochastic dynamical systems, specifically targeting partial stability and partial-state stabilization. We ensure partial asymptotic stability in probability for the closed-loop nonlinear system using a Lyapunov function that is both positive definite and decrescent with respect to a subset of the system state. This Lyapunov function corresponds to the steady-state solution of the stochastic Hamilton-Jacobi-Bellman equation, thereby establishing both probabilistic partial stability and optimality. The proposed framework lays the underpinning for generalizing linear-quadratic optimal control strategies to more complex settings involving nonlinear, nonquadratic, and partial-state stochastic stabilization problems. We also explore the connection between our approach and optimal regulation in both linear and nonlinear time-varying stochastic systems with quadratic and more general cost functionals. In addition, we develop optimal feedback controllers for affine stochastic nonlinear systems through an inverse optimality approach specifically designed for the partial-state stabilization setting.

Keywords: Stochastic optimal control, Process Control, Markov processes
Abstract: Controlling colloidal self-assembly into specific configurations for multi-functional materials has attracted vast interests among material scientists, control engineers, and computer scientists. While various feedback control strategies have demonstrated efficacy in rapidly steering the colloidal particles into desired configurations, dealing the stochasticity still stands as a challenge. Here we investigate the application of stochastic model predictive control (sMPC) on an electric field-mediated colloidal self-assembly process, simulated by Markov state models. Specifically, we examine the robustness of sMPC with respect to model accuracy, considering the challenge of obtaining accurate process models for a high-dimensional, stochastic many-body assembly system in practice. Our findings indicate that with a terminal reward objective function, the proposed sMPC controller can yield satisfactory results, and the sMPC also tends to hold robustness against modeling error. We anticipate the work presented here to stimulate further investigations on applying sMPC for colloidal and nano-particle self-assembly systems.

Keywords: Stochastic optimal control, Reinforcement learning, Lyapunov methods
Abstract: In this paper, we address the concern of potential changes in the state-input distribution during the policy update step of model-free reinforcement learning (RL). In most RL settings, the control policy is updated based on previously recorded data, and then evaluated over the system's future trajectory. However, the new policy may alter the closed-loop behavior and result in future trajectories that are out of distribution with respect to the past learning data, possibly compromising the system's safety due to this discrepancy. To tackle this, we propose a framework that penalizes distributional shifts during RL execution. We incorporate a regularization term that systematically dampens deviations in the distribution in the state-input space, facilitating incremental exploration. Simulations demonstrate the efficacy of our approach.

Keywords: Fault detection, Autonomous robots, Cooperative control
Abstract: We present a progress-based fault detection module and its integration with dynamic task allocation for heterogeneous robot teams. The detector monitors a normalized task-completion signal with a lightweight Kalman filter (KF) and a normalized innovation squared (NIS) test, augmented with a low-rate stall gate, an uncertainty gate, and debounce logic. Health estimates influence the allocator via health-weighted costs and health-dependent masks; reallocation is event-triggered and regularized with an l1 assignment-change penalty to limit reassignment churn while preserving feasibility through slack variables. The detector has constant per-robot update cost, and the allocation remains a convex quadratic program (QP). Experiments on a common team-task setup evaluate measurement-noise increases, velocity-slip biases, communication dropouts, and task abandonment. The results show timely detection in the noise and bias cases, maintained task completion with limited reassignment, and the expected observability delays under communication dropouts.

Keywords: Fault detection, Estimation, Linear systems
Abstract: A fault estimator is proposed for singular linear systems subject to disturbances. This model accounts for faults that explicitly appear both in the state equation and in the system outputs (i.e., system, actuator and sensor faults). A key aspect of this design is that it does not require estimating either the state vector or the disturbances, which provides greater flexibility in structural conditions compared to the requirements needed to implement an unknown input observer. Furthermore, the number of unknown inputs may exceed the number of available outputs. The faults are formulated through an algebraic expression involving high-order derivatives of the system output. In this way, the recovery of the fault signals is achieved using a high-order sliding mode differentiator, which requires that the derivative of the faults maintains a bounded norm.

Keywords: Fault detection, Fault diagnosis, Closed-loop identification
Abstract: Utilizing relay-induced limit cycles to stimulate system dynamics while preserving closed-loop stability, this paper proposes an innovative fault detection and diagnosis (FDD) architecture for closed-loop control systems. The methodology integrates analytical Describing Function (DF) modeling with Fast Fourier Transform (FFT) to provide a compact, fault-sensitive feature vector capable of detecting minor parametric deviations often masked by feedback dynamics. Fault isolation is achieved by comparing test feature vectors with a database of known fault signatures using weighted cosine similarity (WCS) and a confidence-based classification scheme. This integrated method supports diagnostic and prognostic goals, makes finding early signs of problems easier, and encourages a shift toward maintenance based on conditions. A validation study is performed on a relay-excited DC motor model, demonstrating its suitability for real-time, safety-critical applications while maintaining scalability and interpretability.

Keywords: Fault detection, Fault tolerant systems, Observers for Linear systems
Abstract: Control systems can be severely impacted by node faults due to natural failures or malicious attacks. This paper considers two design problems for unknown input observer (UIO)-based fault detection schemes. First, we consider the problem of choosing optimal detection thresholds to ensure the fault is detected before safety violations occur while avoiding false alarms. We derive bounds on the thresholds to ensure that any constant-magnitude fault can be detected. Second, we investigate the problem of ensuring that the fault detector has adequate information to detect and isolate faults. We derive sufficient conditions for the existence of a UIO as a function of the available sensors, and prove that the problem of selecting a minimum-size set of neighbors is equivalent to matroid intersection, enabling polynomial-time approximation algorithms. Our simulation results confirm that our approach is able to detect faults before safety violations occur.

Keywords: Fault detection, Fault tolerant systems
Abstract: Inertial navigation systems (INS) use measurements from inertial sensors (accelerometers and gyroscopes) to estimate the location of a moving object. The estimated location can be used by a control system to move the object to a desired location. If one of the sensors used by the INS becomes faulty during the system operation, the estimated location becomes inaccurate, which can lead to damages to the system or the environment. Therefore, in safety-critical applications such as aircraft navigation, redundant sensors are used to detect and disconnect faulty sensors, which can be expensive. In this paper, we propose a data-driven, redundancy-free fault-tolerant inertial navigation system based on sensor-to-sensor identification. We identify sensor-to-sensor models between sensors from a single inertial measurements unit (IMU). These models, which are identified under healthy conditions of the INS system, are used along with a subset of sensor measurements to predict the response of the remaining sensor measurements. If a sensor becomes faulty, then we replace its actual faulty measurement with a healthy virtual measurement obtained using the identified sensor-to-sensor model and the remaining sensor measurements. We apply the proposed approach to two experimental setups, namely, a rotating platform and a moving vehicle. The proposed approach can be used in commercial applications to ensure low cost safe operation of the INS system.

Keywords: Fault detection, Optimization, Linear systems
Abstract: This study establishes a connection between the output-to-output gain (OOG), a sensitivity metric quantifying the impact of stealthy attacks,

and a novel input-to-input gain (IIG) introduced to evaluate fault sensitivity under disturbances, and investigates their fundamental performance limitations arising from the transmission zeros of the underlying dynamical system. Inspired by the OOG, which characterizes the maximum performance loss caused by stealthy attacks, the IIG is proposed as a new measure of robust fault sensitivity and is defined as

the maximum energy of undetectable faults for a given disturbance intensity. Then, using right (for OOG) and left (for IIG) co-prime factorizations, both metrics are expressed as the H infinity norm of a ratio of the numerator factors. This unified representation facilitates

a systematic analysis of their fundamental limitations. Subsequently, by utilizing the Poisson integral relation, theoretical bounds for the IIG and OOG are derived, explicitly characterizing their fundamental limitations imposed by system non-minimum phase (NMP) zeros. Finally, a

numerical example is employed to validate the results.

Keywords: Optimal control, Optimization algorithms, Computational methods
Abstract: This paper offers a unified perspective on different approaches to the solution of optimal control problems through the lens of constrained sequential quadratic programming. In particular, it allows us to find the relationships between Newton’s method, the iterative LQR (iLQR), and Differential Dynamic Programming (DDP) approaches to solve the problem. It is shown that the iLQR is a principled SQP approach, rather than simply an approximation of DDP by neglecting the Hessian terms, to solve optimal control problems that can be guaranteed to always produce a cost-descent direction and converge to an optimum; while Newton’s approach or DDP do not have similar guarantees, especially far from an optimum. Our empirical evaluations on the pendulum and cart-pole swing-up tasks serve to corroborate the SQP-based analysis proposed in this paper.

Keywords: Optimal control, Optimization algorithms, Predictive control for nonlinear systems
Abstract: Modern Optical Intersatellite Links require sub-degree pointing accuracy at a target satellite moving relative to the ego-spacecraft. This precise pointing presents significant challenges for spacecraft attitude control systems. This paper develops a novel nonlinear optimization algorithm that avoids linearizing the dynamics on the SO(3) manifold, exploiting the causal structure of the problem using block Gauss-Seidel iterations and Riccati iterations of the state and co-state to achieve optimal reference tracking of a moving reference. The proposed algorithm achieves linear complexity in the planning horizon and maintains exact SO(3) manifold constraints throughout the optimization process. Simulations demonstrate superior tracking performance compared to a linearized attitude planning, while maintaining feasible computation speed. These results demonstrate that our algorithm provides a viable solution for next-generation spacecraft attitude control systems requiring precise reference tracking capabilities.

Keywords: Optimal control, Sampled-data control, Predictive control for linear systems
Abstract: This paper introduces a predictive control barrier function (PCBF) framework for enforcing state constraints in discrete-time systems with unknown relative degree, which can be caused by input delays or unmodeled input dynamics. Existing discrete-time CBF formulations typically require the construction of auxiliary barrier functions when the relative degree is greater than one, which complicates implementation and may yield conservative safe sets. The proposed PCBF framework addresses this challenge by extending the prediction horizon to construct a CBF for an associated system with relative degree one. As a result, the superlevel set of the PCBF coincides with the safe set, simplifying constraint enforcement and eliminating the need for auxiliary functions. The effectiveness of the proposed method is demonstrated on a discrete-time double integrator with input delay and a bicopter system with position constraints.

Keywords: Optimization, Optimal control, Estimation
Abstract: Global optimization has gained attention over the past decades, thanks to the development of both theoretical foundations and efficient numerical routines. Among recent advances, Kernel Sum of Squares (KernelSOS) provides a powerful theoretical framework, combining the expressivity of kernel methods with the guarantees of SOS optimization. In this paper, we take KernelSOS from theory to practice and demonstrate its use on challenging control and robotics problems. We identify and address the practical considerations required to make the method work in applied settings: restarting strategies, systematic calibration of hyperparameters, methods for recovering minimizers, and the combination with fast local solvers. As a proof of concept, the application of KernelSOS to robot localization highlights its competitiveness with existing SOS approaches that rely on heuristics and handcrafted reformulations to render the problem polynomial. Even in the high-dimensional, non-parametric setting of trajectory optimization with simulators treated as black boxes, we demonstrate how KernelSOS can be combined with fast local solvers to uncover higher-quality solutions without compromising overall runtimes.

Keywords: Optimization, Optimal control, Spacecraft control
Abstract: Aerobraking maneuvers use atmospheric drag over multiple atmospheric passes to gradually decelerate a spacecraft and reduce its orbital energy for an eventual capture into a low-altitude orbit. This paper proposes a thrust-augmentation strategy closer to the periapsis to enforce dynamic pressure and heating constraints. We present dynamic pressure-based heuristic throttle laws within a per-pass optimization framework to compute thrust augmentation needed alongside apoapsis trim maneuvers for accomplishing the aerobraking maneuver. Numerical simulations for a representative Mars aerobraking scenario and comparison of the heuristics with a per-pass optimization framework demonstrate that a ramp-down throttle strategy at the beginning of periapsis passage can yield good quality solution with reduced computational time compared to optimized throttle levels.

Keywords: Optimization, Queueing systems, Robotics
Abstract: We study online task allocation for multi-robot, multi-queue systems with stochastic arrivals at discrete locations and switching delays. We formulate the problem in discrete time; each location can host at most one robot per time slot, servicing a task consumes one slot, switching between locations incurs a one-slot travel delay, and arrivals are independent Bernoulli processes. We formulate a discounted-cost Markov decision process and propose Exhaustive-Serve-Longest (ESL), a simple real-time policy that serves exhaustively when the current location is nonempty and, when idle, switches to a longest unoccupied nonempty location. We prove the optimality of such policy under symmetry conditions on the arrival processes at each location. We compare in simulation the performance of the ESL policy with two well-known policies: a first-come, first-serve policy and a fixed-dwell cyclic policy. Across variations in server–location ratios and loads, ESL consistently yields lower discounted holding cost and smaller mean queue lengths, with action-time fractions showing more serving and restrained switching. When the symmetry conditions are violated, we compare the performance of the ESL policy on smaller problems with an optimal dynamic programming solution, and determine that the ESL performance is within 2% of the optimal performance. Its simplicity and robustness make ESL a practical default for real-time multi-robot scheduling systems.

Keywords: Building and facility automation, Grey-box modeling, Predictive control for linear systems
Abstract: This paper presents a tube-based model predictive control (Tube-MPC) framework for multi-zone heating, ventilation, and air conditioning (HVAC) systems in airport terminals. The proposed method explicitly addresses system uncertainties arising from passenger flow fluctuations and weather variations, ensuring robust satisfaction of thermal comfort and operational constraints. By incorporating disturbance-invariant sets into the control design, the controller maintains feasible operation while improving resilience to external perturbations. Simulation studies on a five-zone airport terminal case demonstrate that, compared with conventional MPC, the proposed Tube-MPC approach provides stronger disturbance rejection in the hall zone while maintaining the comfort range.

Keywords: Predictive control for nonlinear systems, Hierarchical control, Machine learning
Abstract: Today's control systems are often characterized by modularity and safety requirements to handle complexity, resulting in the use of hierarchical control structures. Although hierarchical model predictive control offers favorable properties, achieving a provably safe, yet modular design remains a challenge. This paper introduces a contract-based hierarchical control strategy to improve the performance of control systems facing challenges related to model inconsistency and independent controller design across hierarchies. We consider a setup where a higher-level controller generates references that affect the constraints of a lower-level controller, which is based on a soft-constrained MPC formulation. The optimal slack variables of the lower-level MPC serve as the basis for a contract that allows the higher-level controller to assess the feasibility of the reference trajectory without exact knowledge of the model, constraints, and cost of the lower-level controller. To ensure computational efficiency while maintaining model confidentiality, we propose using an explicit function approximation, such as a neural network, to represent the cost of optimal slack values. The approach is tested for a hierarchical control setup consisting of a planner and a motion controller as commonly found in autonomous driving.

Keywords: Predictive control for nonlinear systems, Machine learning, Neural networks
Abstract: Model Predictive Control (MPC) offers rigorous safety and performance guarantees but is computationally intensive. Approximate MPC aims to circumvent this drawback by learning a computationally cheaper surrogate policy. Common approaches focus on imitation learning (IL) via behavioral cloning (BC). However, BC fundamentally fails to provide accurate approximations when MPC solutions are set-valued due to non-convex constraints or local minima. We propose a two-stage IL procedure to accurately approximate nonlinear, potentially set-valued MPC policies. The method integrates an approximation of the MPC's optimal value function into a one-step look-ahead loss function, and thereby embeds the MPC's constraint and performance objectives into the IL objective. We prove statistical consistency for policies that exactly minimize our IL objective, implying convergence to a safe and stabilizing control law, and establish input-to-state stability guarantees for approximate minimizers. Simulations demonstrate improved performance compared to BC.

Keywords: Predictive control for nonlinear systems, Modeling, Control applications
Abstract: Active cell balancing (ACB) is an important paradigm in an electric vehicle’s (EV) battery pack. However, in designing the ACB network (ACBN), the interconnections among cells and the selection of power electronic components lead to large variations. Therefore, to evaluate the performance of these diversifications, a modular and scalable framework is required. Consequently, this work utilizes graph theory to model and evaluate the ACBN performance for EV range extension. In this respect, a buck-boost (BB) converter-based architecture, which comprises N series adjacent cells, is modeled using the multiplex (M)–a variant of a graph. Moreover, the high-fidelity expressions for balancing currents are integrated with M. Building on that, a nonlinear model predictive control (NMPC) is formulated and subsequently solved to perform ACB. Similarly, the stability argument for the NMPC based on N cells is provided, and the provision to perform ACB for various ACBN types under the graph-NMPC-based framework is discussed. The simulations are performed for various reallife driving scenarios, and the results demonstrate that the BB converter-based architecture has been able to extend the average EV range up to 32 km under NMPC-based balancing.

Keywords: Predictive control for nonlinear systems, Optimal control, Modeling
Abstract: Multimodal legged-aerial robots offer unique advantages in navigating complex environments by seamlessly transitioning between terrestrial and aerial locomotion modes. However, coordinating propulsive forces from thrusters with ground contact forces presents significant control challenges. This paper presents a contact-implicit model predictive control(CI-MPC) framework that unifies the control of legged and aerial subsystems without predefined contact schedules. We formulate the problem using a 2D thruster-assisted hopper as a reduced-order model, embedding Moreau time-stepping scheme within the MPC dynamics constraints to handle contact forces through differential inclusion and complementarity conditions. This formulation allows ground reaction forces to emerge naturally from the physical constraints. Simulation results demonstrate the framework’s capability to traverse diverse terrains including multi-level steps, uneven surfaces, and wave-like obstacles while maintaining a forward velocity of 1 m/s.Analysis reveals that the contact constraints lead to physically consistent behaviors, with ground reaction forces residing on friction cone boundaries. This unified framework eliminates mode-switching discontinuities and enables the natural discovery of hybrid locomotion strategies, offering a promising direction for controlling multimodal robotic systems.

Keywords: Robust control, Linear parameter-varying systems, Autonomous systems
Abstract: Attitude stabilization of unmanned aerial vehicles (UAVs) in uncertain environments presents significant challenges due to nonlinear dynamics, parameter variations, and sensor limitations. This paper presents a comparative study of H∞ and classical PID controllers for multi-rotor attitude regulation in the presence of wind disturbances and gyroscope noise. The flight dynamics are modeled using a linear parameter-varying (LPV) framework, where nonlinearities and parameter variations are systematically represented as structured uncertainties within a linear fractional transformation formulation. A robust controller based on H∞ formulation is designed using only gyroscope measurements to ensure guaranteed performance bounds. Nonlinear simulation results demonstrate the effectiveness of the robust controllers compared to classical PID control, showing significant improvement in attitude regulation under severe wind disturbances.

Keywords: Robust control, Mechanical systems/robotics, Uncertain systems
Abstract: This paper introduces a new formulation of the disturbance observer (DOB) for a class of uncertain one degree-of-freedom mechanical systems from the perspective of Lagrangian mechanics. Unlike most existing works where the inverse of the nominal model is explicitly used to estimate the lumped disturbance, we derive an alternative expression of the lumped disturbance by differentiating the Lagrangian (i.e., the difference between the kinetic and potential energies), which yields an inverse-model-free DOB structure. The proposed inverse-model-free approach requires less model information for DOB design (e.g., the centripetal/Coriolis term). While our new formulation inherently introduces a singularity issue, we show that such singularities can be handled using a damped pseudo-inverse. Using singular perturbation theory, we rigorously show that this preserves the desired practical tracking behavior under the stated conditions of the DOB compared with other inverse-model-based designs. Simulation results also show that the damped pseudo-inverse provides additional freedom to handle measurement noise, allowing the proposed DOB to outperform existing methods in noisy environments.

Keywords: Robust control, Optimal control, Uncertain systems
Abstract: This paper proposes a robust regret control framework in which the performance baseline adapts to the realization of system uncertainty. The plant is modeled as a discrete-time, uncertain linear time-invariant system with real-parametric uncertainty. The performance baseline is the optimal non-causal controller constructed with full knowledge of the disturbance and the specific realization of the uncertain plant. We show that a controller achieves robust additive regret relative to this baseline if and only if it satisfies a related, robust H_infty performance condition on a modified plant. One technical issue is that the modified plant can, in general, have a complicated nonlinear dependence on the uncertainty. We use a linear approximation step so that the robust additive regret condition can be recast as a standard mu-synthesis problem. A numerical example is used to demonstrate the proposed approach.

Keywords: Robust control, Power electronics, LMIs
Abstract: This paper introduces a new robust multi-objective control strategy for DC-DC converters under input saturation and parametric uncertainties. An integral state-feedback controller is designed using linear matrix inequalities not only to ensure precise reference track- ing, but also to ensure that control signals remain within defined bounds based on a deadzone nonlinearity and a sector condition-based design. The controller also effectively rejects disturbances using an H∞ approach. Additionally, parametric uncertainties of the converter are incorporated into the controller design. The performance of the proposed controller is evaluated under various scenarios using a real prototype. The experimental results demonstrate the effectiveness and robustness of the proposed approach

Keywords: Robust control, Power electronics
Abstract: When electric vehicles (EVs) are integrated into the standalone DC microgrids (DCMGs), stability issues arise due to their constant power load (CPL) behavior, which introduces negative incremental impedance (NII). In addition, the DCMGs suffer from an inherent low-inertia problem. Therefore, this study presents a composite controller incorporating a global integral terminal sliding mode controller with a backstepping controller. A virtual capacitor is employed to mitigate the lower inertia issue and strengthen the DC-bus voltage response. An improved fractional power-based reaching law decreases chattering as well as accelerates convergence. Exact feedback linearization converts the nonlinear power converter model into Brunovsky's canonical form, thereby mitigating NII effects and non-minimum phase issues. The entire system stability is verified using Lyapunov control theory. Finally, simulation outcomes confirm the superior performance of the proposed approach, with 34.4-53.3% reduction in overshoot, 52.9-74.9% in undershoot, and 12-47.4% in settling time compared to the existing controller.

Keywords: Robust control, Robotics, Lyapunov methods
Abstract: This work presents a robust backstepping type controller formulation for the position tracking control of engineering systems actuated via electro-hydraulic actuators (EHAs). Specifically, a robust controller that does not require accurate knowledge of the system parameters and uses only position measurements is proposed. A filtered based approach is applied to remove the velocity dependency of the controller formulation. Stability of the closed loop system and the uniform boundedness of the tracking error signals are ensured via Lyapunov based arguments. The overall performance of the proposed method is illustrated, initially through physics-based MATLAB/Simscape studies, and then experimentally on a 1 degree of freedom (dof) EHA test-bed and a 2 dof robotic arm.

Keywords: Modeling, Smart structures, Robotics
Abstract: Evolved from the twisted string actuator, the coiled string actuator (CSA) produces appreciable and reasonably consistent actuation by coiling a pair of fully twisted strings. Being a recently discovered compliant actuator, only the CSA with a varying twisting and coiling zone has been considered, namely, the strings are coiled as a continuum. CSA in this configuration cannot be easily applied to applications where coiled strings have direct contact with the system. It is strongly desirable to study CSA with a constant twisting and coiling zone, whose string sliding region can be conveniently embedded into systems like tendons. CSA in this configuration exhibits a unique simultaneous twisting-induced and coiling-induced actuation, which challenges both the actuator design and modeling. In this study, we propose a separator structure design that decreases the friction caused by the string-separator contact. A kinematic model is proposed that captures the simultaneous twisting-induced and coiling-induced actuation of the CSA with a constant twisting and coiling zone. Encouraging preliminary experimental characterizations and model simulation results are provided.

Keywords: Control applications, Robotics, Intelligent systems
Abstract: Soft robots offer significant advantages in safety and adaptability, yet achieving precise and dynamic control remains a major challenge due to their inherently complex and nonlinear dynamics. Recently, Data-enabled Predictive Control (DeePC) has emerged as a promising model-free method that bypasses explicit system identification by directly leveraging input–output data. While DeePC has shown success in other domains, its application to soft robots remains underexplored, particularly for 3D soft robotic systems. This paper addresses this gap by developing and experimentally validating an effective DeePC framework on a 3D cable-driven soft arm. Specifically, we design and fabricate a soft robotic arm with a thick tubing backbone for stability, a dense silicone body with large cavities for strength and flexibility, and rigid endcaps for secure termination. Using this platform, we implement DeePC with singular value decomposition (SVD)-based dimension reduction for two key control tasks: fixed-point regulation and trajectory tracking in 3D space. Comparative experiments with a baseline model-based controller demonstrate DeePC’s superior accuracy and adaptability, highlighting its potential as a practical solution for dynamic control of soft robots.

Keywords: Control applications, Robotics, Reinforcement learning
Abstract: Soft snake robots offer remarkable flexibility and adaptability in complex environments, yet their control remains challenging due to highly nonlinear dynamics. Existing model-based and bio-inspired controllers rely on simplified assumptions that limit their performance. Deep reinforcement learning (DRL) has recently emerged as a promising alternative, but online training is often impractical because of costly and potentially damaging real-world interactions. Offline RL provides a safer option by leveraging pre-collected datasets, but it suffers from distribution shift, which degrades generalization to unseen scenarios. To overcome this challenge, we propose DiSA-IQL (Distribution-Shift-Aware Implicit Q-Learning), an extension of IQL that incorporates robustness modulation by penalizing unreliable state–action pairs to mitigate distribution shift. We evaluate DiSA-IQL on goal-reaching tasks across two settings: in-distribution and out-of-distribution evaluation. Simulation results show that DiSA-IQL consistently outperforms baseline models, including Behavior Cloning (BC), Conservative Q-Learning (CQL), and vanilla IQL, achieving higher success rates, smoother trajectories, and greater robustness.

Keywords: Autonomous robots, Reinforcement learning, Adaptive systems
Abstract: Soft robotic crawlers are mobile robots that utilize soft body deformability and compliance to achieve locomotion through surface contact. Designing control strategies for such systems is challenging due to model inaccuracies, sensor noise, and the need to discover locomotor gaits. In this work, we present a model-based reinforcement learning (MB-RL) framework in which Gaussian latent dynamics inferred from onboard sensors serve as a predictive model that guides an actor-critic algorithm to optimize locomotor policies. We evaluate the framework on a minimal crawler model in simulation using inertial measurement units and time-of-flight sensors as observations. The learned latent dynamics enable short-horizon motion prediction while the actor-critic discovers effective locomotor policies. This approach highlights the potential of latent-dynamics MB-RL for enabling embodied soft robotic adaptive locomotion based solely on noisy sensor feedback.

Keywords: Robotics, Modeling, Model Validation
Abstract: Pneumatic Artificial Muscles (PAMs) are a type of soft actuator used in robotics to mimic the behavior of biological muscle, typically consisting of an inner bladder and a form of outer sleeve that causes contraction when pressurized. Many mathematical models have been developed to predict the force and displacement of PAMs. However, these models are mainly sensitive to the variation in materials and geometries of PAMs, making them limited in practical use for specific PAMs. This paper presents a unified physics-informed geometric model of PAMs for a range of sizes with initial lengths of 60, 70, and 80 mm and initial diameters ranging from 3.175, 6.35, and 9.525 mm. The fabrication and experimental testing of a type of PAM produced in our Lab for the muscle-driven snake-like robot locomotion is discussed. A series of experimental isometric and isotonic tests was conducted on customized PAMs with varying geometries made of the same materials. Finally, an empirical modeling approach was utilized, which resulted in a geometric model that preserves the structure of the Pujana-Aresse model. The model is limited to the size range tested in the lab and has only been validated for the aforementioned sizes. These muscle sizes were explored based on the overall geometry of the snake and the desired application for the muscles. For validation, the model prediction was compared to the experimental results, yielding a mean Normalized Root Mean Square Error (NRMSE) of 16.3% for the isometric force and 12% for the isotonic displacement.

Keywords: Robotics, Adaptive systems, Control applications
Abstract: Nasal swabbing is an essential procedure for detecting various respiratory diseases, and performing it with a robot can reduce the risk of infection for medical personnel. However, nasal swab robots face several challenges. For example, the narrow and elongated nasal passage allows only a small pose error during swab insertion, and the internal nasal structure varies slightly depending on each individual’s health condition. To deal with these problems, this paper proposes an admittance control law that enables the robot to adjust the swab’s pose based on the feedback of contact force and torque. However, the choice of admittance parameters significantly affects the behavior of the robot. To reduce the contact force and alleviate oscillation of the swab simultaneously, an adaptive admittance parameter strategy is introduced that modifies the admittance parameter in real time according to the pose of the swab. Consequently, safety, efficiency, and comfort of the nasal swabbing process are enhanced. Then, the proposed adaptive admittance control law was implemented in a 6-axis industrial robot, and a curved channel was constructed as a test environment for experimental verification. The results showed that a satisfactory balance among all performance indices, regarding safety, efficiency, and comfort, was achieved.

Keywords: Quantum information and control, Computational methods, Optimization
Abstract: This paper explores the integration of Carleman linearization and quantum annealing to solve nonlinear ordinary differential equations (ODEs) under the Chebyshev spectral method. Using Carleman linearization, the computation of optimal Chebyshev coefficients is reformulated as a quadratic program, which can be further transformed into a Quadratic Unconstrained Binary Optimization (QUBO) problem via binary expansion. This QUBO formulation is well-suited for quantum annealers, where solutions are obtained by minimizing the Hamiltonian of the corresponding Ising model. To address hardware limitations, we employ an iterative quantum annealing strategy with dynamically updated parameters. Case studies on the D-Wave platform demonstrate the potential of our approach to harness the strengths of Carleman linearization and Chebyshev approximation theory, offering continuous and differentiable solutions to nonlinear ODEs, especially promising when quantum resources become more economically viable.

Keywords: Quantum information and control, Identification
Abstract: Recent advancements in quantum computing, secure communication, and precision sensors indicate a transition from theoretical exploration to practical applications in quantum technologies. This paper addresses state estimation and control of superconducting flux qubits, a promising platform for scalable quantum computation. A novel adaptive sliding mode observer (ASMO) is proposed for simultaneous robust state reconstruction and online estimation of unknown damping parameters β and Λ in the presence of dissipation and measurement noise. Using incomplete measurement data, the observer ensures accurate estimation while maintaining robustness to model uncertainties. A sliding mode control law is then developed to stabilize the system and achieve desired trajectory tracking. Lyapunov-based analysis guarantees convergence of the estimation and control errors and ensures closed-loop stability. Simulation results demonstrate the effectiveness and robustness of the proposed framework, indicating its potential to enhance the reliability and stability of superconducting quantum processors under realistic operating conditions

Keywords: Quantum information and control, Machine learning, Neural networks
Abstract: This paper studies temporal learning with quantum reservoirs driven by continuous-time dynamics. Our goal is to model long-range dependencies under irregular sampling while avoiding backpropagation through time and repeated numerical ODE integration. We formulate a Continuous-time Quantum Reservoir (CQR) framework in which an input-modulated Hamiltonian drives Schrödinger evolution, and a classical linear readout is fitted from sampled quantum features. We analyze three variants: (i) CQR-unitary, which uses expectation values of Hermitian observables; (ii) CQR-measurement, which uses projective-measurement probabilities; and (iii) CQR-trainable, which additionally tunes input-coupling Hamiltonian parameters via parameter-shift updates. Experiments on Lorenz forecasting, human-activity classification, and two regression benchmarks show competitive accuracy, improved memory capacity, and clear compute-accuracy trade-offs across variants. We also provide a Stone-Weierstrass-based universality result for the induced feature algebra and discuss practical design limitations.

Keywords: Game theory, Quantum information and control, Optimization
Abstract: In this paper, we develop a framework for deception in quantum games, extending the Honey-X paradigm from classical zero-sum settings into the quantum domain. Building on a view of deception in classical games as manipulation of a player’s perception of the payoff matrix, we formalize quantum deception as controlled perturbations of the payoff Hamiltonian subject to a deception budget. We show that when victims are aware of possible deception, their equilibrium strategies surprisingly coincide with those of naive victims who fully trust the deceptive Hamiltonian. This equivalence allows us to cast quantum deception as a bilevel optimization problem, which can be reformulated into a bilinear semidefinite program. To illustrate the framework, we present simulations on quantum versions of the Penny Flip game, demonstrating how quantum strategy spaces and non-classical payoffs can amplify the impact of deception relative to classical formulations.

Keywords: Quantum information and control, Optimal control, Machine learning
Abstract: In this paper, we propose a new framework for quantum optimal actuator selection that unifies control allocation with optimal quantum control design. To guarantee the uniqueness of solutions, we transform the infinite-horizon control allocation problem into an equivalent finite-horizon setting. Leveraging the isometry of the Bloch representation, we systematically characterize available actuators and introduce a quantitative measure of controllability based on accessible control degrees of freedom. Building on this foundation, we develop a model-free, trajectory-informed machine learning approach that evaluates control performance directly from system trajectories to identify the optimal actuator set. Finally, we provide rigorous analysis demonstrating that, even under unknown system dynamics, the proposed method converges to the optimal value function for a sufficiently large time horizon.

Keywords: Quantum information and control
Abstract: We propose a quantum-assisted framework for solving constrained finite-horizon nonlinear optimal control problems using a barrier Sequential Quadratic Programming (SQP) approach. A quantum subroutine is incorporated to efficiently solve the Schur complement step using block-encoding and Quantum Singular Value Transformation techniques. We formally analyze the time complexity and convergence behavior under the cumulative effect of quantum errors, establishing local input-to-state stability and convergence to a neighborhood of the stationary point, with explicit error bounds in terms of the barrier parameter and quantum solver accuracy. The proposed framework enables computational complexity to scale polylogarithmically with the system dimension demonstrating the potential of quantum algorithms to enhance classical optimization routines in nonlinear control applications.

Keywords: Human-in-the-loop control, Automotive control, Cooperative control
Abstract: Driving style is a key factor in the comfort and acceptance of automated vehicle (AV) features. In SAE Level-2 automation, where the driver must supervise the system and remain ready to intervene, mismatches between the automation’s driving style and the driver’s preference can reduce trust and trigger takeovers. This paper proposes an adaptive driving-style control framework that minimizes such preference mismatch. In a driving-simulator study, we compare fixed, trust-based, and preference-based adaptation heuristics and analyze their effects on preference mismatch and trust. We then train a driving-preference prediction model and use it in an implicit adaptation policy that selects among bounded driving styles for upcoming events. A validation study shows that the predictive policy achieves equal or lower preference mismatch than comparison baselines, particularly when starting from a less defensive style, while also yielding higher average trust. The results provide a step toward developing human-aware driving automation that can implicitly adapt its driving style to the driver’s preferences.

Keywords: Human-in-the-loop control, Intelligent systems, Control applications
Abstract: Industry 4.0 in pharmaceutical manufacturing demands advanced automation and control to handle increasingly specialized drugs and personalized therapies. However, leveraging sophisticated controls like model predictive control (MPC) often requires expertise beyond the typical plant operator. This paper proposes a human-in-the-loop framework where a large language model (LLM) assists operators in configuring and tuning an MPC using natural language commands. The LLM translates high-level objectives (e.g. “speed up the response,” “avoid overshoot”) into MPC tuning adjustments and provides real-time analysis of controller performance. A human supervisor remains in the loop as a safety net, which is essential in regulated pharma environments to ensure trust and compliance. The concept is demonstrated on a Temperature Control Lab (TCLab) hardware setup using a GEKKO-based MPC. Results show that key control objectives can be modified on the fly via natural language, enabling intuitive operator interaction. Four representative scenarios illustrate how LLM-translated commands affect MPC behavior, and an insight-generation feature provides automated textual analysis of performance. This human-in-the-loop, LLM-assisted MPC approach offers a pathway to deploy advanced control in pharma manufacturing while preserving human oversight.

Keywords: Human-in-the-loop control, Learning, Optimal control
Abstract: This paper introduces an energy-efficient distributed optimal control framework for human–multi-robot cooperative manipulation, combining a spiking neural network (SNN) observer for biologically inspired intent estimation with a game-theoretic distributed optimal control strategy for coordination. At the estimation level, the event-driven SNN captures human motor intent from local force feedback and consensus information, enabling real-time trajectory estimation through sparse spike processing. Unlike conventional neural networks, the SNN processes asynchronous events to significantly reduce computational energy, while singular value decomposition (SVD)–based weight tuning laws ensure stable gradient propagation during online learning.

At the control level, a multilayer NN-based actor-critic architecture applies adaptive dynamic programming within a cooperative game-theoretic setting. Coupled Hamilton-Jacobi-Bellman equations are solved through neighborhood optimization, allowing each robot to minimize a performance cost that incorporates local dynamics, neighboring interactions, and human-robot force coordination. Simulation results in human–multi-robot collaborative tasks, achieve superior tracking accuracy with a 45% reduction in operational cost compared to state-of-the-art methods. The SNN-based observer further reduces energy consumption by 60% while maintaining estimation accuracy, enabling practical real-time deployment on resource-constrained robotic systems.

Keywords: Human-in-the-loop control, Modeling
Abstract: This paper quantifies the performance of human-autonomy teams (HATs) by measuring the distance between observed strategies and the generalized Nash equilibrium (GNE). Traditional approaches of assessing HATs do not fully capture the bidirectional interactions or provide actionable insight to improve team behavior. Here, we model HATs as non-zero-sum games and introduce a potential function (informed by those seen in potential games) to characterize how a HAT's behavior diverges from the identified GNE strategy. In a user study with a shared-control vehicle simulator, we apply the metric to assess the effectiveness of the teams formed between an autonomous agent and human participants. We further identify distinct behavior modes of participants based on the trends in the potential function values across different ratios of shared control. The results demonstrate the utility of the potential function in distinguishing patterns of human behavior, providing a pathway towards enhancing the effectiveness of HATs.

Keywords: Human-in-the-loop control
Abstract: By combining automation accuracy with human adaptability, shared control provides enhanced performance and safety in dynamic, complex environments. Traditional arbitration methods for integrating automation and human inputs often rely on system-specific, parameter-dependent functions that are based on shared control metrics such as trust, workload, or attention. Meanwhile, Control Barrier Functions (CBFs) enforce safety constraints on automated systems but are typically limited to safeguarding plant states. This work introduces a novel arbitration method based on Control Barrier Functions (CBFs), where shared control metrics such as workload, attention, and trust are expressed as real-time inequality constraints. The resulting quadratic-programming formulation determines the automation assistance input that enforces these constraints while preserving feasibility and safety. This CBF-based arbitration provides a systematic, interpretable, and scalable foundation for safe human–autonomy integration.

Keywords: Autonomous systems, Constrained control, Human-in-the-loop control
Abstract: Shared autonomy envisions seamless machine assistance to human drivers, enhancing both driving safety and personalization of the driving experience. In this paper, we propose a shared control algorithm for a lane-keeping application that formally guarantees the vehicle remains within a predefined track while minimizing intrusiveness. This is accomplished through the development of a family of Control Barrier Functions that construct a nested sequence of safe subsets, each representing a different level of conservativeness. The level of conservativeness is then adapted online to individual drivers based on their inferred, uncertain driving performance, using a Bayesian filter. We employ a gradient-based algorithm to automatically optimize the hyperparameters, thereby minimizing the intrusiveness of each Control Barrier Functions in parallel. The proposed approach achieves an average computation time of 0.0047 seconds, and its effectiveness is validated through comprehensive simulation studies and real-world hardware experiments with a human driver in the loop.

Keywords: Optimization algorithms, Network analysis and control
Abstract: Many combinatorial optimization problems (COPs) are NP-hard and, in general, hard to solve. Over the past years, many unconventional computing paradigms have been proposed for solving such problems. Notably, Ising machines are one such paradigm for solving COPs with binary variables, which have attracted sustained attention in recent years. Specifically, Ising machines are a class of Langevin models for sampling low-energy configurations of the Ising model. Since many COPs such as the max-cut problem can be directly mapped to the Ising model, Ising machines solve the COPs by finding or sampling lowest-energy configurations (ground states) of the Ising model. Although Ising machines have shown great potential, they remain somewhat inadequate in addressing multi-state COPs. Here, inspired by the 3-state clock model, a generalization of the Ising model to a multi-state model, we introduce a nonlinear dynamical model, referred to as 3-state clock machine, for solving max-3-cut problems.

Keywords: Network analysis and control, Control of networks, Large-scale systems
Abstract: We present the Social Influence Game (SIG), a framework for modeling adversarial persuasion in social networks with an arbitrary number of competing players. Our goal is to provide a tractable and interpretable model of contested influence that scales to large systems while capturing the structural leverage points of networks. Each player allocates influence from a fixed budget to steer opinions that evolve under DeGroot dynamics, and we prove that the resulting best-response optimization problem is a difference-of-convex program. To enable scalability, we develop an Iterated Linear (IL) solver that approximates player objectives with linear programs. In experiments on random and archetypical networks, IL achieves solutions within 7% of nonlinear solvers while being over 10× faster, scaling to large social networks. This paper lays a foundation for asymptotic analysis of contested influence in complex networks.

Keywords: Network analysis and control, Cooperative control
Abstract: This paper studies an algorithm that generates generalized lollipop graphs for any feasible pair of numbers of vertices and edges. An explicit expression for the total effective resistance of these graphs is derived, and it is shown that this value achieves or approximates the maximal possible total effective resistance among all connected graphs with the same vertex and edge counts. In particular, the gap between the total effective resistance of the constructed generalized lollipop graphs and the extremal value converges to zero as the graph size tends to infinity.

Keywords: Network analysis and control, Optimization, Sensor networks
Abstract: We provide a distributed online algorithm for multi-agent submodular maximization under communication delays. We are motivated by the future distributed information-gathering tasks in unknown and dynamic environments, where utility functions naturally exhibit the diminishing-returns property, i.e., submodularity. Existing approaches for online submodular maximization either rely on sequential multi-hop communication, resulting in prohibitive delays and restrictive connectivity assumptions, or restrict each agent’s coordination to its one-hop neighborhood only, thereby limiting the coordination performance. To address the issue, we provide the Distributed Online Greedy (DOG) algorithm, which integrates tools from adversarial bandit learning with delayed feedback to enable simultaneous decision-making across arbitrary network topologies. We provide the approximation performance of DOG against an optimal solution, capturing the suboptimality cost due to decentralization as a function of the network structure. Our analyses further reveal a trade-off between coordination performance and convergence time, determined by the magnitude of communication delays. By this trade-off, DOG spans the spectrum between the state-of-the-art fully centralized online coordination approach [1] and fully decentralized one-hop coordination approach [2].

Keywords: Network analysis and control
Abstract: The convergence of opinions in the Friedkin-Johnsen (FJ) framework is well studied, but the topological conditions leading to opinion clustering remain less explored. To bridge this gap, we examine the role of topology in the emergence of opinion clusters within the network. The key contribution of the paper lies in the introduction of the notion of topologically prominent agents, referred to as Locally Topologically Persuasive (LTP) agents. Interestingly, each LTP agent is associated with a unique set of (non-influential) agents in its vicinity. Using them, we present conditions to obtain opinion clusters in the FJ framework in any arbitrarily connected digraph. A key advantage of the proposed result is that the resulting opinion clusters are independent of the edge weights and the stubbornness of the agents. Finally, we demonstrate using simulation results that, by suitably placing LTP agents, one can design networks that achieve any desired opinion clustering.

Keywords: Machine learning, Network analysis and control, Reinforcement learning
Abstract: In many real-world applications such as recommendation systems, multiple learning agents must balance exploration and exploitation while maintaining safety guarantees to avoid catastrophic failures. We study the stochastic linear bandit problem in a multi-agent networked setting where agents must satisfy stage-wise conservative constraints. A network of N agents collaboratively maximizes cumulative reward while ensuring that the expected reward at every round is no less than (1-alpha) times that of a baseline policy. Each agent observes local rewards with unknown parameters, but the network optimizes for the global parameter (average of local parameters). Agents communicate only with immediate neighbors, and each communication round incurs additional regret. We propose MA-SCLUCB (Multi-Agent Stage-wise Conservative Linear UCB), an episodic algorithm alternating between action selection and consensus-building phases. We prove that MA-SCLUCB achieves regret tilde{O}left(frac{d}{sqrt{N}}sqrt{T}cdotfrac{log(N T)}{sqrt{log(1/|lambda_2|)}}right) with high probability, where d is the dimension, T is the horizon, and |lambda_2| is the network's second largest eigenvalue magnitude. Our analysis shows: (i) collaboration yields frac{1}{sqrt{N}} improvement despite local communication, (ii) communication overhead grows only logarithmically for well-connected networks, and (iii) stage-wise safety adds only lower-order regret. Thus, distributed learning with safety guarantees achieves near-optimal performance in reasonably connected networks.

Keywords: Traffic control, Constrained control, Feedback linearization
Abstract: This paper presents a novel control framework that integrates Filtered Feedback Linearization (FFL) with Model Predictive Control (MPC) to manage traffic flow in a homogeneous highway network. FFL is a high-gain robust control method that relies on minimal knowledge of the traffic model and can effectively reject unknown and unmeasured disturbances; however, it does not inherently handle actuator constraints, which may lead to control saturation. Conversely, MPC effectively enforces constraints and optimizes control actions, but its performance depends on model accuracy and it lacks inherent robustness to disturbances. To bridge these limitations, we propose an FFL-MPC framework in which FFL ensures disturbance rejection while MPC regulates virtual control inputs within feasible limits. A key contribution of this work is a novel constraint mapping algorithm that dynamically transforms actuator constraints into the virtual control domain, preserving compatibility with the low-pass filtering dynamics of FFL. The METANET model is employed to model the traffic dynamics. The proposed framework is validated through numerical simulations of highway traffic, demonstrating its effectiveness in alleviating congestion, enforcing control limits, and preserving robust performance under model uncertainty and external disturbances.

Keywords: Simulation, Modeling, Estimation
Abstract: Urban traffic anomalies, such as collisions and disruptions, threaten the safety, efficiency, and sustainability of transportation systems. In this paper, we present a simulation-based framework for modeling, detecting, and predicting such anomalies in urban networks. Using the Simulation of Urban MObility (SUMO) platform, we generate reproducible rear-end and intersection crash scenarios with matched baselines, enabling controlled experimentation and comparative evaluation. We record vehicle-level travel time, speed, and emissions for both edge- and network-level analysis. Building on this dataset, we develop a hybrid forecasting architecture that combines bidirectional long short-term memory networks with a diffusion convolutional recurrent neural network to capture temporal dynamics and spatial dependencies. Our simulation studies on the Broadway corridor in New York City demonstrate the framework's ability to reproduce consistent incident conditions, quantify their effects, and provide accurate multi-horizon traffic forecasts. Our results highlight the value of combining controlled anomaly generation with deep predictive models to support reproducible evaluation and sustainable traffic management.

Keywords: Traffic control, Multivehicle systems, Transportation networks
Abstract: In this paper, we aim to mitigate congestion in traffic management systems by guiding travelers along system-optimal (SO) routes. However, we recognize that most theoretical approaches assume perfect driver compliance, which often does not reflect reality, as drivers tend to deviate from recommendations to fulfill their personal objectives. Therefore, we propose a route recommendation framework that explicitly learns partial driver compliance and optimizes traffic flow under realistic adherence. We first compute an SO edge flow through flow optimization techniques. Next, we train a compliance model based on historical driver decisions to capture individual responses to our recommendations. Finally, we formulate a stochastic optimization problem that minimizes the gap between the target SO flow and the realized flow under conditions of imperfect adherence. Our simulations conducted on a grid network reveal that our approach significantly reduces travel time compared to baseline strategies, demonstrating the practical advantage of incorporating learned compliance into traffic management.

Keywords: Traffic control, Sampled-data control, Autonomous vehicles
Abstract: This paper addresses the enforcement of Disturbance String Stability (DSS) in connected autonomous vehicle platoons subject to asynchronous communication, sampling, and quantization in case of a constant time headway spacing policy. Building on a mesoscopic control framework, we design digital controllers that guarantee DSS despite external disturbances and device limitations. Simulation results demonstrate the effectiveness of the proposed approach.

Keywords: Transportation networks, Traffic control, Emerging control applications
Abstract: This paper presents a framework connecting nonlinear consensus dynamics with traffic assignment problems in multi-route transportation networks. We define a model of route-choice behavior in which flow evolution is governed by first-in-first-out dynamics, with route costs combining baseline and congestion terms. This formulation yields Wardrop equilibria as consensus states, where all used routes have equalized costs. The analysis establishes key system properties, including flow conservation, invariance of unused routes, and uniqueness of the symmetric equilibrium. Stability is demonstrated through monotonic convergence, while case studies of balanced, congested, and asymmetric networks illustrate flow reallocation and steady-state concentration on optimal routes. This framework provides a foundation for decentralized route selection in intelligent transportation systems and enables control strategies that exploit consensus-like rerouting in urban traffic.

Keywords: Transportation networks, Traffic control, Lyapunov methods
Abstract: This work presents a dynamic pricing scheme which guarantees global asymptotic stability of the Nash equilibrium set for a discrete-time routing game. We adopt a game theoretic approach in which each population consists of commuters between a particular origin-destination pair. Populations revise their strategies according to payoff discrepancies between routes, where the strategy revision process is modeled by the class of impartial pairwise comparison (IPC) evolutionary dynamic models (EDMs). We show that the EDM satisfies a discrete-time dissipativity property and leverage this insight to design a dynamic payoff mechanism acting as a supplemental toll imposed on each population. By introducing an augmented storage function which governs the toll dynamics, we ensure the control input (i.e., toll) asymptotically vanishes at the Nash equilibrium set, thus guaranteeing convergence of the closed-loop system. We validate our findings with numerical experiments.

Keywords: Decentralized control, Multivehicle systems, Autonomous robots
Abstract: Existing decentralized methods for multi-agent motion planning lack formal, infinite-horizon safety guarantees, especially for communication-constrained systems. We present R3R which, to our knowledge, is the first decentralized and asynchronous framework for multi-agent motion planning under range-limited communication constraints with infinite-horizon safety guarantees for systems of nonlinear agents. R3R's novelty lies in combining our gatekeeper safety framework with a geometric constraint termed R-Boundedness, which together establish a formal link between an agent's communication radius and its ability to plan safely. We constrain trajectories to lie within a fixed planning radius, determined by a function of the agent's communication radius. This enables trajectories to be certified as provably safe for all time using only local information. Our algorithm is fully asynchronous, and ensures the forward invariance of these guarantees even in time-varying networks where agents asynchronously join and replan. We evaluate our approach in simulations of up to 128 Dubins vehicles, validating our theoretical safety guarantees in dense, obstacle-rich scenarios. We further show that R3R's computational complexity scales with local agent density rather than problem size, providing a practical solution for scalable and provably safe multi-agent systems.

Keywords: Multivehicle systems, Cooperative control, PID control
Abstract: This paper proposes a novel distributed proportional-integral-derivative (PID) controller design using the dominant pole-placement method for homogeneous vehicle platoon systems with directed and undirected topologies under the constant time headway policy (CTHP). Analytical expressions for PID gains are derived using a coefficient matching approach, considering all real non-dominant poles. String stability criteria are derived for both the directed predecessor-following (PF) and undirected bidirectional (BD) topologies. A genetic algorithm (GA) is employed to optimize three key elements: (i) closed-loop design parameters, (ii) PID controller gains, and (iii) time headway. The algorithm searches the stabilizable region of the controller parameter space to satisfy derived PID analytical expressions and string stability criteria while minimizing the integral square error (ISE). The proposed PID controller ensures both internal and string stability. The effectiveness of the design is validated through numerical simulations on a five-vehicle platoon with PF and BD topologies under CTHP.

Keywords: Multivehicle systems, Lyapunov methods, Autonomous systems
Abstract: In this paper, a Lyapunov-based framework for flexible, distance-based, and leaderless formation control of multi-agent systems (MAS) in space-restricted and obstacle-populated environments is proposed. A control Lyapunov function methodology that considers a generic dynamic model of autonomous vehicles is introduced to derive decentralized control laws that are computed locally by each agent and guarantee a number of crucial properties. Among others, the feedback laws ensure that the swarm satisfies strict prespecified formation bounds, while guaranteeing inter-agent collision avoidance, convergence of the vehicles’ velocities to the desired values and adherence to hard upper and lower bounds, and satisfaction of spatial restrictions imposed by the navigation environment. Simulation results on a swarm of fixed-wing unmanned aerial vehicles validate the theoretical findings.

Keywords: Sensor fusion, Multivehicle systems, Kalman filtering
Abstract: A multi-robot, target tracking algorithm is proposed that improves tracking accuracy by imposing field-of-view (FOV) constraints based on each robot’s sensor characteristics. The algorithm is independent of the underlying tracking scheme and is evaluated using experimental data collected from sentry-type robots whose trajectories are independent of positive detections of the target. These data are processed using an unscented Kalman filter (UKF) with and without the FOV constraint. The UKF with FOV constraint delivers significantly less tracking error and recovers much more quickly than the standard UKF in periods where the target is unobserved.

Keywords: Cooperative control, Traffic control, Simulation
Abstract: In this paper, we present a hierarchical framework integrating upper-level routing with low-level optimal trajectory planning for connected and automated vehicles (CAVs) in urban networks. The upper-level controller distributes traffic flows using a dynamic re-routing algorithm that leverages real-time density information and fundamental diagrams to predict when edges reach critical density, proactively adjusting routing weights to prevent congestion. The low-level controller coordinates CAVs at signal-free intersections, generating optimal, fuel-efficient trajectories while ensuring safety through constraint satisfaction. We formulate this as an optimal control problem with an analytical solution. SUMO micro-simulation experiments on a realistic network demonstrate that our framework significantly outperforms baseline static routing, achieving 55.8% reduction in delays and 12.1% reduction in travel times, showing substantial improvements in urban traffic efficiency

Keywords: Transportation networks, Autonomous systems, Cooperative control
Abstract: This paper proposes a coordinated routing approach that investigates the use of connected and automated vehicles (CAVs) in dedicated bus lanes. The aim is to improve bus schedule adherence while enhancing the travel efficiency of CAVs during the transitional phase of mixed traffic environments. Our approach utilizes real-time traffic data to dynamically reroute CAVs in anticipation of congestion. By continuously monitoring traffic conditions on dedicated lanes and tracking the real-time positions of buses, the system adjusts CAV routes in advance to avoid potential interference with operating buses. This cooperation reduces CAV travel times and minimizes delays that impact transit services. The proposed strategy is validated using microscopic traffic simulations in SUMO. The results demonstrate significant improvements in both transit on-time performance and CAV travel efficiency across a range of traffic conditions.

Keywords: Energy systems, Constrained control, Supervisory control
Abstract: Hybrid power plants (HPPs) combine multiple generation sources and energy storage to support generation shortfalls and grid-service demands. This paper introduces a modeling and control design framework for wind-solar-battery HPPs. Specifically, this work adapts established modeling paradigms for wind farms, solar plants and battery models into a supervisory-control-oriented form. In the case of wind and battery models, generator torque and cell current control laws are developed using nonlinear control and control barrier function techniques to track commands issued by a supervisory controller while maintaining safe and stable operation. The framework is demonstrated on a utility demand-tracking task under time-varying wind and irradiance using a rule-based supervisory control law.

Keywords: Energy systems, Optimization algorithms, Power systems
Abstract: By providing the optimal operating point that satisfies both the power flow equations and engineering limits, the optimal power flow (OPF) problem is central to the operation of electric power systems. While extensive research efforts have focused on reliably computing high-quality OPF solutions, assessing the feasibility of transitioning between operating points remains challenging since the feasible spaces of OPF problems may consist of multiple disconnected components. It is not possible to transition between operating points in different disconnected components without violating OPF constraints. To identify such situations, this paper introduces an algorithm for certifying the infeasibility of transitioning between two operating points within an OPF feasible space. As an indication of potential disconnectedness, the algorithm first seeks an infeasible point on the line connecting a pair of feasible points. The algorithm then certifies disconnectedness by using convex relaxation and bound tightening techniques to show that all points on the plane that is normal to this line are infeasible. Using this algorithm, we provide the first certifications of disconnected feasible spaces for a variety of OPF test cases.

Keywords: Energy systems, Power systems, Optimization
Abstract: This paper examines the problem of maximizing the endurance of an unmanned aerial vehicle. The paper focuses on a tiltrotor vehicle configuration. This configuration has the advantage of combining vertical takeoff/landing with more efficient horizontal flight. However, this comes at the disadvantage of a substantial power demand discrepancy between takeoff/landing versus cruise. This discrepancy motivates powertrain hybridization. Specifically, the paper examines a parallel hybrid powertrain that combines the energy and power density advantages of internal combustion and battery-electric propulsion, respectively, using a mix of lithium-ion and lithium-sulfur batteries. We define stochastic endurance in terms of the ability to execute a Monte Carlo-simulated family of flight missions without power shortages. We then pose a control co design problem, where the objective is endurance maximization and the optimization variables are the sizing of the powertrain components and the parameterization of an onboard power management policy. This policy uses Gaussian kernels to compute the power split ratio between the powertrain subsystems based on total power demand and the “states of charge” of the onboard energy stores. A simulation study shows that this approach leads to attractive endurance levels and supports mission planning by quantifying the impact of initial fuel and payload allocations on endurance.

Keywords: Building and facility automation, Energy systems, Linear systems
Abstract: How much energy, money, and emissions can advanced control of heating and cooling equipment save in real buildings? To address this question, researchers sometimes control a small number of thermal zones within a larger multi-zone building, then report savings for the controlled zones only. That approach can overestimate savings by neglecting heat transfer between controlled zones and adjacent zones. This paper mathematically characterizes the overestimation error when the dynamics are linear and the objectives are linear in the thermal load, as usually holds when optimizing energy efficiency, energy costs, or emissions. Overestimation errors can be large even in seemingly innocuous situations. For example, when controlling only interior zones that have no direct thermal contact with the outdoors, all perceived savings are fictitious. This paper provides an alternative estimation method based on the controlled and adjacent zones' temperature measurements. The new method does not require estimating how much energy the building would have used under baseline operations, so it removes the additional measurement and verification challenge of accurate baseline estimation.

Keywords: Energy systems, Uncertain systems, Optimal control
Abstract: Large aggregations of distributed energy resources (DERs) are increasingly used to provide grid services, such as ramping and regulation, yet their integration is challenged by volatile net demand and stringent real-time safety requirements. This work proposes a risk-sensitive closed-loop model predictive control (MPC) framework to optimally allocate heterogeneous DERs while mitigating the impact of rare but severe demand excursions, such as sudden load spikes of large electric loads. The MPC approach leverages the fact that short-term load forecasts are more accurate, enabling each step in the rolling optimization sequence to use refined predictions while ensuring constraint satisfaction. To achieve this, the closed-loop model-predictive controller is parametrized using the System-Level Synthesis (SLS) framework, and risk aversion is achieved by embedding a Conditional Value-at-Risk (CVaR) constraint that explicitly limits the expected impact of extreme forecast errors. Numerical simulations on demand-forecast datasets demonstrate the proposed approach's ability to reduce the power required from bulk generation in real time and to account for forecast demand uncertainty.

Keywords: Building and facility automation, Energy systems, Mechatronics
Abstract: Space conditioning accounts for nearly half of building energy use and associated CO_2 emissions. Conventional indoor environmental control strategies rely on heating, ventilation, and air conditioning systems to regulate indoor temperature reliably, but they often do not exploit passive strategies such as shading and natural ventilation, which limits potential energy savings. Purely passive mechanisms, on the other hand, cannot maintain comfort across all climates and seasons. This paper develops a coordinated strategy for the joint control of active heating and cooling systems and operable passive elements. The proposed strategy combines decentralized PI control of active heating and cooling, with reinforcement learning (RL) agents that actuate operable passive elements such as window openings. The control objective is to minimize energy consumption while keeping zone temperatures within a prescribed band. Because the active and passive inputs are coupled through thermal dynamics, analytical conditions for Lyapunov stability and ultimate boundedness are derived for this joint control strategy. The portability of the proposed approach was demonstrated by training the RL agent on a single-zone building and deploying it on a multi-zone building without retraining. EnergyPlus simulations of a multi-zone dwelling in three cities and two seasons show reductions of 40--74% in degree-hour discomfort and 54--86% in active energy loads relative to a baseline PI controller with rule-based logic for passive elements. The coordinated controller kept the indoor temperature within 0.5^circC below and 1.4^circC above the comfort band. These results highlight the potential of coordinated active and passive strategies to deliver substantial energy savings while maintaining occupant comfort.

Keywords: Chemical process control, Machine learning, Sensor networks
Abstract: Integrated process systems enhance operational efficiency but increase the cyberattack surface due to dense interconnections and reliance on networked sensing. This study develops a community-aware distributed transformer framework that jointly performs attack detection and localization of compromised sensors. A spectral community detection algorithm is used to decompose the graph representation of the process into strongly interacting communities. Within each community, a transformer learns the dynamics of local measured outputs and identifies sensor-level attack likelihoods and a community-level detection flag. System-level decisions are then obtained by aggregating the outputs across communities. The framework is evaluated on an integrated benzene alkylation process, with training data generated from closed-loop simulations under model predictive control and moving horizon estimation across multiple adversarial scenarios, including single- and multi-sensor attacks. Compared with centralized end-to-end and two-tier transformer baselines, the distributed approach attains higher accuracy, F1 scores, and Jaccard indices, demonstrating improved robustness and more precise localization of compromised sensors.

Keywords: Chemical process control, Process Control, Machine learning
Abstract: In this paper, we propose symbolic decision trees as surrogate models for approximating model predictive control laws. The proposed approach learns simultaneously the partition of the input domain (splitting logic) as well as local nonlinear expressions for predicting the control action leading to interpretable piecewise nonlinear control laws. The local nonlinear expressions are determined by the learning problem and are modeled using a set of basis functions. The learning task is posed as a mixed integer optimization problem, which is solved to global optimality with state-of-the-art global optimization solvers. We apply the proposed approach to a case study regarding the control of an isothermal reactor. The results show that the proposed approach can learn the control law accurately, leading to closed-loop performance comparable to that of a standard model predictive controller. Finally, comparison with existing interpretable models shows that the symbolic trees achieve both lower prediction error and superior closed-loop performance.

Keywords: Chemical process control, Process Control, Machine learning
Abstract: This paper addresses the solution of nonlinear dynamic optimization problems that compute optimal manipu- lated input profiles to enforce desired output profiles. Such tra- jectory optimization problems commonly arise in chemical pro- cess applications, for example, batch processes where optimal temperature or feeding profiles (in case of fed-batch processes) are calculated to enforce time-varying product quality profiles, tightly controlling the reaction rate or rate of heat generation. We propose deep neural operators that approximate function to function mappings as surrogates for the solution of such dynamic optimization problems. We specifically employ deep operator networks (DeepONets) and Fourier-enhanced Deep- ONets in a batch polymerization reactor case study for which number-average and weight-average molecular weight profiles, together with a final conversion target, are enforced through an optimal temperature program. Our results show that the Fourier-enhanced DeepONet architecture performs very well in approximating the solution of the dynamic optimization problem for different instances, achieving a lower prediction error compared to the standard DeepONet architecture and standard feedforward neural networks.

Keywords: Process Control, Optimization, Optimal control
Abstract: Industrial refrigeration systems have tremendous energy needs, but optimizing their operation remains challenging due to the tension between minimizing energy costs and meeting strict cooling requirements. Load shifting--strategic overcooling in anticipation of future demands--offers substantial efficiency gains. This work seeks to rigorously quantify these potential savings through the derivation of optimal load shifting policies. Our first contribution establishes a novel connection between industrial refrigeration and inventory control problems with convex ordering costs, where the convexity arises from the fundamental relationship between energy consumption and cooling capacity. Leveraging this formulation, we derive three main theoretical results: (1) an optimal algorithm for deterministic demand scenarios, with proof that optimal trajectories are non-increasing which is a valuable structural insight for practical control; (2) performance bounds that quantify the value of load shifting as a function of cost convexity, demand variability, and temporal patterns; (3) a computationally tractable load shifting heuristic with provable near-optimal performance under uncertainty. Numerical simulations validate our theoretical findings, and a case study using real industrial refrigeration data demonstrates an opportunity for improved load shifting.

Keywords: Chemical process control, Process Control, Pattern recognition and classification
Abstract: Alarm floods are critical operational challenges in process industries, overwhelming operators and increasing the risk of missed or delayed responses. This paper presents a context-enriched Word2Vec framework for real-time alarm flood classification, designed to capture temporal spacing, activation order, and alarm priority within an embedding space. By extending the generic Skip-gram model with enriched tokenization and priority-weighted training, the method addresses key challenges such as variable sequence lengths, overlapping alarms across categories, and imbalance between high- and low-priority events. The framework is evaluated on the Tennessee Eastman benchmark, where it achieved high classification accuracy, with early divergence of similarity trajectories enabling timely identification of fault categories. Compared to a generic Word2Vec baseline, the enriched embeddings consistently improved both early-stage and overall classification performance while maintaining computational efficiency suitable for online applications. These results highlight the potential of computationally efficient natural language processing (NLP)-inspired approaches to support operators with faster and more reliable decision-making during industrial alarm floods.

Keywords: PID control, Reinforcement learning, Decentralized control
Abstract: Designing controllers for industrial processes is a daunting task, particularly because most industrial processes are multi-input multi-output (MIMO) in nature. This introduces coupling effects and loop interactions, which are not encountered in the case of single-input single-output (SISO) processes. In this manuscript, a novel multi-agent reinforcement learning-based approach for designing adaptive PI controllers is proposed. The Smith predictor is incorporated to compensate for the deadtime in the diagonal elements of the process transfer function. The proposed method employs the Deep Deterministic Policy Gradient (DDPG) algorithm with Long-Short-Term Memory (LSTM) based actor-critic network architecture to design adaptive PI controllers for a MIMO process. The performance of the proposed controller is measured against some of the existing methods for designing classical PI controllers, using performance indices such as Integral Absolute Error (IAE) and Integral Square Error (ISE). The proposed controller demonstrates superior performance compared to classical state-of-the-art PI controllers. The proposed controller is performing well even in the presence of plant-model mismatches, delay uncertainties, and disturbances, demonstrating robust performance.

Keywords: Robotics, Control applications, Cooperative control
Abstract: Utilizing teams of multiple robots is advantageous for handling bulky objects. Many related works focus on multi-manipulator systems, which are limited by workspace constraints. In this paper, we extend a classical hybrid motion-force controller to a team of legged manipulator systems, enabling collaborative loco-manipulation of rigid objects with a force-closed grasp. Our novel approach allows the robots to flexibly coordinate their movements, achieving efficient and stable object co-manipulation and transport, validated through extensive simulations and real-world experiments.

Keywords: Robotics, Optimal control, Autonomous systems
Abstract: We present a method for learning unknown parametric constraints from locally-optimal input-output trajectory data. We assume the data is generated by rollouts of stochastic nonlinear dynamics, under a single state or output feedback law and initial condition but distinct noise realizations, to robustly satisfy underlying constraints despite worst-case noise outcomes. We encode the Karush-Kuhn-Tucker (KKT) conditions of this robust optimal feedback control problem within a feasibility problem to recover constraints consistent with the local optimality of the demonstrations. We prove that our constraint learning method (i) accurately recovers the demonstrator’s policy, and (ii) conservatively estimates the set of policies that ensure constraint satisfaction despite worst-case noise realizations. Moreover, we perform sensitivity analysis, proving that when demonstrations are corrupted by transmission error, the inaccuracy in the learned feedback law scales linearly in the error magnitude. Empirically, our method accurately recovers unknown constraints from simulated noisy, closed-loop demonstrations generated using dynamics, both linear and nonlinear, (e.g., unicycle and quadrotor) and a range of feedback mechanisms.

Keywords: Robotics, Reinforcement learning, Autonomous robots
Abstract: The objective of this study is to develop a model-free taskspace trajectory planner for space manipulators that operates in real time, using a Twin Delayed Deep Deterministic Policy Gradient (TD3) agent to enable safe and reliable debris capture. A local control strategy with singularity avoidance and manipulability enhancement is employed to ensure stable execution. The manipulator must simultaneously track a capture point on a non-cooperative target, avoid self-collisions, and prevent unintended contact with the target. To address these challenges, we propose a curriculum-based multi-critic network where one critic emphasizes accurate tracking and the other enforces collision avoidance. A prioritized experience replay buffer is also used to accelerate convergence and improve policy robustness. The framework is evaluated on a simulated seven-degree-of-freedom KUKA LBR iiwa mounted on a free-floating base in Matlab/Simulink, demonstrating safe and adaptive trajectory generation for debris removal missions.

Keywords: Vision-based control, Robust control
Abstract: Perception-based neural network controllers are increasingly used in autonomous systems that rely on visual inputs to operate in the real world. Ensuring the safety of such systems under uncertainty is challenging. Existing verification techniques typically focus on Lp-bounded perturbations in the pixel space, which fails to capture the low-dimensional structure of many real-world effects. In this work, we introduce a novel verification framework for perception-based controllers that can generate outer-approximations of reachable sets through explicitly modeling uncertain observations with geometric perturbations. Our approach constructs a boundable mapping from states to images, enabling the use of state-based verification tools while accounting for uncertainty in perception. We provide theoretical guarantees on the soundness of our method and demonstrate its effectiveness across benchmark control environments. This work provides a principled framework for certifying the safety of perception-driven control systems under realistic visual perturbations.