This paper proposes a resilient distributed event-triggered (ET) control framework for nonlinear networked control systems (NCSs) modelled via the Takagi-Sugeno (T-S) fuzzy approach under stochastic cyber-attacks. Unlike conventional centralised ET strategies, the proposed scheme adopts a fully decentralised triggering mechanism in which each sensor independently determines its transmission instants. This design not only alleviates communication congestion and reduces energy consumption but also provides scalability to large-scale interconnected systems. To accurately capture adversarial behaviours, a novel modelling framework is developed that simultaneously incorporates two distinct classes of nonlinear cyber-attacks reflecting more realistic and compounded threat scenarios. By embedding these attacks into the control design, the proposed scheme strengthens system resilience against compounded adversarial scenarios. Rigorous stability conditions are derived through Lyapunov theory and formulated in terms of linear matrix inequalities (LMIs), ensuring stability of the closed-loop system. The effectiveness of the proposed framework is validated on a mass spring system through both numerical simulations and real-time experimental tests, demonstrating its ability to preserve stability and security of the closed-loop system under adverse cyber-attack scenarios.

In this work, we propose an adaptive stepsize rule with guaranteed convergence for Distributed Gradient Tracking applied to scalar quadratic problems with heterogeneous curvatures. Most distributed gradient-based algorithms require a suitable stepsize selection. Available theoretical bounds are often overly conservative, while practical implementations typically rely on empirically tuned heuristics. Online adaptive strategies have only recently emerged for general distributed convex optimization, but their properties and performance remain only partially understood. To gain analytical insight, we focus on the informative setting of scalar quadratic costs, which allows us to explicitly capture the interplay between network topology and curvature heterogeneity. We derive a convergence bound parameterized only by the essential spectral radius of the consensus matrix and the heterogeneity of the local cost curvatures, both computable online without any prior knowledge of the optimization problem. Optimizing this bound yields a computationally tractable surrogate for the convergence rate and the optimal constant stepsize. The resulting stepsize admits an analytical interpretation, guarantees convergence for arbitrary network topologies and curvature heterogeneity, and is provably tight for complete graphs and homogeneous curvatures. Finally, extensive numerical simulations demonstrate that the proposed distributed adaptive strategy significantly outperforms existing offline and online stepsize selection rules in the considered setting.

Timescale separation is a powerful tool for analyzing interconnected dynamical systems. Meanwhile, operator theory provides a general framework for studying the convergence of iterative methods, including algorithms in optimization, learning, and control. In this paper, we bridge these two areas by establishing timescale separation results for a class of interconnected fixed-point iterations. As is customary in timescale separation, our results involve auxiliary systems that separately capture the dynamics induced by the slow and fast operators, arising from the original interconnection in the limit as the timescale parameter tends to zero. In particular, by assuming contractivity and paracontractivity of the respective auxiliary systems, we derive explicit bounds on the timescale parameter that guarantee linear convergence of the original, interconnected system. To illustrate the applicability of our result, we employ it to prove the convergence of a feedback optimization scheme.

Silicon carbide (SiC) power modules are increasingly deployed in automotive traction inverters, where condition monitoring is essential to prevent in-service failures. Despite extensive qualification under AQG 324, no consolidated approach exists for in-field health state estimation: physics-of-failure lifetime models lack real-time applicability, purely data-driven architectures require large labeled datasets and generalize poorly, and physics-informed frameworks remain too demanding for embedded deployment. We address SiC MOSFET modules assembled with sintered packaging, which suppresses solder degradation and produces aging behavior distinct from previously studied devices. Instead of the smooth quasi-exponential drift of solder-based modules, the forward voltage drop V_DS exhibits multi-regime profiles, with wirebond liftoff events introducing abrupt, non-monotonic perturbations. We propose a condition monitoring framework combining three elements. First, physics-informed features replace raw sensor signals with cumulative damage indicators derived from junction temperature swing, mean junction temperature and a Miner rule accumulator, encoding degradation history in an interpretable form. Second, a monotonicity constraint enforced by gradient penalty regularization embeds the expected degradation direction as a physics-guided prior. Third, a heavy-tailed output distribution replaces the point estimate, giving calibrated uncertainty robust to the out-of-distribution variance introduced by liftoff. On an industrial power cycling dataset from Infineon Technologies, several neural architectures are compared under a strict cross-validation protocol. The full configuration reduces mean absolute error by approximately 70

To achieve optimal power control of an electric vehicle (EV) charging station, this article presents a two-stage model predictive control (MPC) approach considering environmental and economic impacts. The charging station that includes a battery energy storage system (BESS) and a photovoltaic (PV) system is also connected to the main power grid. The proposed approach includes two stages: the first stage is introduced to reduce carbon dioxide (CO2) emissions and costs associated with interaction with the main power grid. This stage itself consists of two sublevels: in the first sublevel, an offline single-step optimization problem is solved based on expected data for the entire day, while the second sublevel is designed to reduce the impact of data uncertainties used in the previous step. Then, a real-time event-triggered MPC (ETMPC) is configured based on the new constraints of the power to be purchased from the main power grid, resulting from the first stage. The proposed scheme considers EV energy demands and departure times, and by leveraging vehicle-to-grid (V2G) capabilities and the constraints obtained from the first stage, it achieves more efficient power control. To reduce the computations needed to solve the online optimization problem at each time step, new conditions based on an event-driven approach are introduced. Also, to improve the feasibility of the ETMPC, an ellipsoidal terminal constraint is introduced. In addition, due to the event-triggered mechanism, new conditions are presented for proving the descent of the cost function, which guarantees the stability of the closed-loop system. Finally, various numerical simulations and experimental tests are carried out in order to highlight the environmental, economic, and computational impacts of the proposed method.

Timescale separation is a powerful tool for analyzing interconnected dynamical systems. Meanwhile, operator theory provides a general framework for studying the convergence of iterative methods formulated as fixed-point iterations, including algorithms arising in optimization, learning, and control. In this paper, we bridge these two areas by establishing timescale separation results for fixed-point iterations induced by both deterministic and stochastic operators. As customary in timescale separation, our results involve auxiliary systems that arise from the original interconnection in the limit as the timescale parameter tends to zero and separately capture the dynamics induced by the slow and fast operators. The proposed operator-theoretic framework yields explicit and readily checkable bounds on this tunable parameter, expressed in terms of standard operator constants. To illustrate the applicability of our results, we employ them to prove the convergence properties of a feedback optimization scheme in both deterministic and stochastic settings.

Vision-language models (VLMs) have recently demonstrated strong capabilities in mapping multimodal observations to robot behaviors. However, most current approaches rely on end-to-end visuomotor policies that remain opaque and difficult to analyze, limiting their use in safety-critical robotic applications. In contrast, classical robotic systems often rely on structured policy representations that provide interpretability, modularity, and reactive execution. This work investigates how foundation models can be specialized to generate structured robot policies grounded in multimodal perception, bridging high-dimensional learning and symbolic control. We propose a neuro-symbolic approach in which a VLM synthesizes executable Behavior Tree policies from visual observations, natural language instructions, and structured system specifications. To enable scalable supervision without manual annotation, we introduce an automated pipeline that generates a synthetic multimodal dataset of domain-randomized scenes paired with instruction-policy examples produced by a foundation model. Real-world experiments on two robotic manipulators show that structured policies learned entirely from synthetic supervision transfer successfully to physical systems. The results indicate that foundation models can be adapted to produce interpretable and structured robot policies, providing an alternative to opaque end-to-end approaches for multimodal robot decision making.

This paper studies multi-agent reinforcement learning with submodular team utilities for online distributed task allocation. In this setting, each agent selects one action from a local categorical policy, so feasible joint actions form a partition matroid over agent-action pairs. Classical multilinear extensions use independent Bernoulli sampling and therefore do not match the categorical policies executed by decentralized agents. To address this mismatch, we introduce the Partition Multilinear Extension (PME), a continuous relaxation whose value equals the expected team utility under factorized categorical policies. We prove that submodular difference rewards provide unbiased PME marginal-gradient information and yield a stagewise score-function policy-gradient estimator. Based on this connection, we propose SubMAPG, a centralized-training decentralized-execution policy-gradient framework with masked categorical policies and submodular difference-reward training signals. For the associated PME marginal-space projected stochastic-gradient dynamics, we prove a stagewise 1/2-approximation guarantee and sublinear dynamic regret in slowly varying environments, measured by the path length of the optimal PME marginals. To handle open systems with time-varying agents and targets, we instantiate SubMAPG with graph neural network policies. Experiments on multi-robot coverage and multi-target tracking show that SubMAPG outperforms local greedy and shared-reward baselines and is competitive with centralized myopic greedy strategies.

This paper studies policy learning for distributed task allocation in open multi-agent systems, where agents may join and leave in a time-varying fashion, with submodular stage team utilities. At each time, the active agents select actions from local categorical policies such that the feasible joint agent-action pairs form a partition matroid. Standard continuous relaxations of submodular set functions are based on independent Bernoulli sampling, making them inconsistent with agents' policies.To solve this mismatch, we propose the partition multilinear extension (PME), a policy-based relaxation whose continuous support matches feasible actions under categorical policies.We prove that the marginal gains of the stage utility provide an unbiased estimator of the gradient of the PME and that maximizing the PME over action distributions is equivalent to maximizing the stage utilities over agent actions, which are critical to devise principled policy gradient.Building on this, we design SubMAPL, a centralized-training decentralized-execution KL-mirror policy-learning method that uses local marginal gains as stochastic PME gradients during training. KL-mirror updates preserve categorical feasibility without Euclidean projection.In the case where agents run tabular-softmax policies, we introduce open policy migration and an open-system KL tracking variation to handle agent arrivals and departures. Using dynamic regret analysis, we establish a lower bound on the cumulative utility which accounts for the openness of the environment and for the gap between optimal stage-wise and global utilities. Simulations on multi-agent coverage demonstrate that SubMAPL outperforms policy-gradient and online-learning baselines.

This paper addresses autonomous intervention with an underwater vehicle–manipulator system (UVMS) in confined, cluttered, and partially known environments, where poor maneuverability, narrow passages, and uncertain execution may cause the robot to enter unrecoverable regions. We propose MANTA, a three-layer hierarchical planning-and-control framework that couples passage accessibility, manipulation feasibility, and closed-loop execution. The first layer performs global connectivity reasoning in a conservative reduced base space to extract traversable corridor candidates toward the task region. The second layer refines each candidate corridor by jointly optimizing the continuous base motion and arm trajectory, producing a collision-free base–arm trajectory. The third layer learns a reach-and-hold base policy using Gaussian-process model-based reinforcement learning (MBRL) through MC-PILCO, enabling trajectory tracking and station keeping at the planned manipulation state. During execution, the framework monitors map updates and can trigger recovery and route repair when the active passage becomes infeasible. MANTA is evaluated in confined UVMS planning and closed-loop tracking experiments. Across 120 matched planning queries, it achieves higher task success than full-state sampling-based baselines while producing larger clearance margins and lower arm motion. The learned MC-PILCO policy further reduces position and yaw tracking errors on both training and unseen tube-like references. These results show MANTA as a structured and data-efficient framework for safe autonomous underwater intervention in caves, tubes, and cluttered subsea structures.

In this paper, we address the problem of achieving fast convergence in gradient descent for quadratic functions without relying on a priori knowledge of global function parameters. Inspired by adaptive stepsize algorithms for smooth convex functions, we propose a computationally lightweight strategy based on running estimates of minimal and maximal local curvatures. We prove that our proposed algorithm converges to the optimal constant stepsize which achieves the fastest convergence. Simulations show that the convergence rate achieved by our proposed algorithm is comparable or superior to recent adaptive approaches both in the quadratic case under consideration and in a preliminary test on logistic classification.

In robotics, many different approaches ranging from classical planning over optimal control to reinforcement learning (RL) are developed and borrowed from other fields to achieve reliable control in diverse tasks. To get a clear understanding of their individual strengths and weaknesses and their applicability in real-world robotic scenarios, it is important to benchmark and compare their performances not only in a simulation but also on real hardware. The second AI Olympics with RealAIGym competition was held at the 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2024) to contribute to this cause and evaluate different controllers according to their ability to solve a dynamic control problem on an underactuated double-pendulum system (Figure 1) with chaotic dynamics. This article describes the four different RL methods submitted by the participating teams; presents their performance in the swing-up task on a real double pendulum, measured against various criteria; and discusses their transferability from simulation to real (sim-to-real) hardware and their robustness to external disturbances.

Buildings consume a large share of global energy, making accurate building energy models (BEMs) critical for improving efficiency. Existing calibration methods often face high computational demands, overlook parameter identifiability, and lack independent validation, limiting their practical reliability. This study proposes a comprehensive framework for the efficient calibration of multi-zone BEMs, integrating graph-based modelling (GBM), global sensitivity analysis (GSA), parameter identifiability assessment, and optimization. GBM captures complex thermal interactions across zones with interpretability and accuracy. Sobol'-based GSA identifies the most influential parameters, improving calibration efficiency, while parameter identifiability assessment ensures uniqueness of parameter estimates before optimization. Calibration is conducted using the golden search optimization algorithm, balancing exploration and exploitation. Validation through both statistical metrics and independent validation data not used during model calibration demonstrates the robustness and generalizability of the proposed approach. Results demonstrate that explicitly addressing sensitivity, identifiability yields more accurate and reliable BEMs, supporting energy-efficient operation and control strategies.HighlightsDeveloping the graph-based model of multi-zone building dynamics with a reduced-complexity structure.Optimized calibration integrating GSA, identifiability, and golden search method. Validation robustness and generalizability under static metrics and independent data not used during model calibration.Providing a foundation method for energy-efficient operation and control strategies.

Data-driven health-state estimators for SiC (Silica-Carbide) power modules typically report their performance on a single accelerated-aging campaign, and how that performance transfers to a different failure mechanism is rarely tested. We benchmark five reference methods from the prognostics and condition-monitoring literature against a physics-informed NODE (Neural Ordinary Differential Equation) on two SiC power-cycling campaigns driven by structurally different failure mechanisms, solder-layer fatigue and wire-bond lift-off, under a per-module k-fold protocol. The NODE is evaluated under two input regimes that share the rest of the pipeline: the baseline electrical precursors and a set of cumulative thermoelectric features. Every reference method degrades on the wire-bond campaign, with average errors growing and precision decreasing with respect to their performance on the soldered campaign. The NODE fed with the cumulative features keeps its soldered-campaign metrics on both mechanisms, with differences inside the fold-to-fold variance, while the same architecture fed with the baseline precursors falls back to the reference-method cluster. The input representation contributes at least as much as the architecture to failure-mechanism transferability of a health-state estimator.

In this work we consider the problem of controlling a dynamical system affected by bounded disturbances and subject to constraints through a wireless network. Existing works guarantee constraint satisfaction despite packet losses and delays by enforcing the past input sequence used in the estimation at the remote unit and the past input sequence applied at the plant to be the same. This condition, typically referred to as prediction consistency, requires strong assumptions on the network, such as a TCP-like protocol or a known delay bound, and conservative control design. In this work, we propose a novel approach to enforce constraints without requiring prediction consistency by handling unmatched input sequences as an additive disturbance. We define a theoretically sound Tube MPC to be implemented at the remote side and we introduce a novel acceptance rule at the plant side that allows to use control packets even if the past input sequences used in the estimation and applied at the plant are different. Simulations with real Wi-Fi channel realizations show that the proposed strategy has a more reactive response and better performance compared to existing solutions.

This short paper describes our proposed solution for the third edition of the "AI Olympics with RealAIGym" competition, held at ICRA 2025. We employed Monte-Carlo Probabilistic Inference for Learning Control (MC-PILCO), an MBRL algorithm recognized for its exceptional data efficiency across various low-dimensional robotic tasks, including cart-pole, ball & plate, and Furuta pendulum systems. MC-PILCO optimizes a system dynamics model using interaction data, enabling policy refinement through simulation rather than direct system data optimization. This approach has proven highly effective in physical systems, offering greater data efficiency than Model-Free (MF) alternatives. Notably, MC-PILCO has previously won the first two editions of this competition, demonstrating its robustness in both simulated and real-world environments. Besides briefly reviewing the algorithm, we discuss the most critical aspects of the MC-PILCO implementation in the tasks at hand: learning a global policy for the pendubot and acrobot systems.

Deep Reinforcement Learning (DRL) has emerged as a powerful paradigm for learning complex policies directly from high-dimensional input spaces, enabling advances across a variety of domains. Modern DRL algorithms often rely on dual-network Q-learning architectures to approximate optimal policies to overcome overestimation bias. Recent research has introduced approaches leveraging multiple Q-functions to further mitigate overestimation effects and enhance policy reliability. However, there is a growing emphasis on deploying DRL in edge scenarios, where privacy concerns and stringent hardware constraints necessitate highly efficient algorithms. In such environments, the computational and memory efficiency of learning methods is of critical importance. In this context, we propose Edge Delayed Deep Deterministic Policy Gradient (EdgeD3), a novel reinforcement learning algorithm specifically designed for edge computing settings. EdgeD3 offers significant reductions in GPU time (by 25%) and computational and memory usage (by 30%), while consistently achieving or surpassing the performance of state-of-the-art algorithms across multiple benchmarks and in real-world tasks. Note to Practitioners- Driven by the growing need for efficient computational solutions in automation, this research introduces the Edge Delayed Deep Deterministic Policy Gradient (EdgeD3), a novel Deep Reinforcement Learning algorithm designed for applications on edge devices with a limited computational budget. EdgeD3 is developed to enable on-device execution of policy learning, when computing resources are at a premium, such as in smart manufacturing and autonomous vehicle systems. EdgeD3 can learn highly effective policies, on par of state-of-the-art algorithms, while utilizing significantly fewer resources. This would empower autonomous devices to operate independently of cloud-based systems, fostering faster operational speeds and enhancing data privacy.

In this paper, we consider a network of agents that jointly aim to minimise the sum of local functions subject to coupling constraints involving all local variables. To solve this problem, we propose a novel solution based on a primal-dual architecture. The algorithm is derived starting from an alternative definition of the Lagrangian function, and its convergence to the optimal solution is proved using recent advanced results in the theory of time-scale separation in nonlinear systems. The rate of convergence is shown to be linear under standard assumptions on the local cost functions. Interestingly, the algorithm is amenable to a direct implementation to deal with asynchronous communication scenarios that may be corrupted by other non-idealities such as packet loss. We numerically test the validity of our approach on a real-world application related to the provision of ancillary services in three-phase low-voltage microgrids.

Deep Reinforcement Learning (RL) has emerged as a powerful method for addressing complex control problems, particularly those involving underactuated robotic systems. However, in some cases, policies may require refinement to achieve optimal performance and robustness aligned with specific task objectives. In this paper, we propose an approach for fine-tuning Deep RL policies using Evolutionary Strategies (ES) to enhance control performance for underactuated robots. Our method involves initially training an RL agent with Soft-Actor Critic (SAC) using a surrogate reward function designed to approximate complex specific scoring metrics. We subsequently refine this learned policy through a zero-order optimization step employing the Separable Natural Evolution Strategy (SNES), directly targeting the original score. Experimental evaluations conducted in the context of the 2nd AI Olympics with RealAIGym at IROS 2024 demonstrate that our evolutionary fine-tuning significantly improves agent performance while maintaining high robustness. The resulting controllers outperform established baselines, achieving competitive scores for the competition tasks.

Underactuated robots play a crucial role in various applications, however, their control is challenging, especially when ensuring compliance with state constraints. This paper proposes to extend the Model-Based RL algorithm Monte Carlo - Probabilistic Inference for Learning Control (MC-PILCO), by incorporating soft constraints. These constraints encourage policy optimization ensuring compliance with state constraints, and enhancing safety during policy execution. We demonstrate the effectiveness of our method on the Acrobot robot, solving the swing-up and stabilization task while adhering to safety limitations. Additionally, we compare MC-PILCO's data efficiency with Model-Free RL algorithms in the simulated Acrobot, highlighting its potential for practical applications. Copyright (c) 2025 The Authors. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/)