Papers

We study state estimation for discrete-time linear stochastic systems under distributional ambiguity in the initial state, process noise, and measurement noise. We propose a noise-centric distributionally robust Kalman filter (DRKF) based on Wasserstein ambiguity sets imposed directly on these distributions. This formulation excludes dynamically unreachable priors and yields a Kalman-type recursion driven by least-favorable covariances computed via semidefinite programs (SDP). In the time-invariant case, the steady-state DRKF is obtained from a single stationary SDP, producing a constant gain with Kalman-level online complexity. We establish the convergence of the DR Riccati covariance iteration to the stationary SDP solution, together with an explicit sufficient condition for a prescribed convergence rate. We further show that the proposed noise-centric model induces a priori spectral bounds on all feasible covariances and a Kalman filter sandwiching property for the DRKF covariances. Finally, we prove that the steady-state error dynamics are Schur stable, and the steady-state DRKF is asymptotically minimax optimal with respect to worst-case mean-square error.

In this paper, we investigate cooperative platoon formation and benefit allocation in mixed-energy truck fleets composed of both electric and fuel-powered trucks. The central challenge arises from the platoon-size constraint, which limits the number of trucks permitted in each platoon and introduces combinatorial coupling into the search for optimal platoon formation structures. We formulate this problem as a coalitional game with bounded coalition sizes and derive a closed-form characterization of the optimal coalition structure that maximizes the fleet-wide platooning benefit. Building on this structure, we develop a type-based least-core payoff allocation scheme that guarantees stability within the coalition-structure core (CS-core). For cases in which the CS-core is empty, we compute the least-core radius to determine the minimal relaxation required to achieve approximate stability. Through numerical studies, we demonstrate that the proposed framework consistently achieves the highest total platooning benefit among all feasible formation configurations while providing stable benefit allocations that outperform existing baseline methods.

In this paper, we study output synchronization for multi-agent systems. The objective is to design a protocol which only depends on the agent dynamics and does not require any knowledge of the network. If the network has a directed spanning tree then the protocols designed in this paper achieve classical output synchronization. Otherwise, the protocol achieves weak synchronization which is induced by network stability in the sense that the signals exchanged over the network converge to zero. Weak sychronization is explained in detail in this paper. Even though we consider linear agents, it is known that this in general requires nonlinear protocols. In the paper we use adaptive protocols. In the literature, two classes of protocols are considered often called collaborative protocols (with additional communication between the protocols and non-collaborative protocols (sometimes referred to as fully decentralized where the additional communication is not present). This paper considers both of these cases.

Unlike traditional multi-agent coordination frameworks, which assume a fixed number of agents, UAS traffic management (UTM) requires a platform that enables Uncrewed Aerial Systems (UAS) to freely enter or exit constrained low-altitude airspace. Consequently, the number of UAS operating in a given region is time-varying, with vehicles dynamically joining or leaving even in dense, obstacle-laden environments. The primary goal of this paper is to develop a computationally efficient management system that maximizes airspace usability while ensuring safety and efficiency. To achieve this, we first introduce physics-informed methods to structure fixed skyroads across multiple altitude layers of urban airspace, with the directionality of each skyroad designed to guarantee full reachability. We then present a novel Continuous UTM (C-UTM) framework that optimally allocates skyroads to UAS requests while accounting for the time-varying capacity of the airspace. Collectively, the proposed model addresses the key challenges of low-altitude UTM by providing a scalable, safe, and efficient solution for urban airspace usability.

Beamspace dimensionality reduction, a classical tool in array processing, has been shown in recent work to significantly reduce computational complexity and training overhead for adaptive reception in massive multiuser (MU) MIMO. For sparse multipath propagation and uniformly spaced antenna arrays, beamspace transformation, or application of a spatial FFT, concentrates the energy of each user into a small number of spatial frequency bins. Empirical evaluations demonstrate the efficacy of linear Minimum Mean Squared Error (LMMSE) detection performed in parallel using a beamspace window of small, fixed size for each user, even as the number of antennas and users scale up, while being robust to moderate variations in the relative powers of the users. In this paper, we develop a fundamental geometric understanding of this “unreasonable effectiveness” in a regime in which zero-forcing solutions do not exist. For simplified channel models, we show that, when we enforce a suitable separation in spatial frequency between users, the interference power falling into a desired user's beamspace window of size $W$ concentrates into a number of dominant eigenmodes smaller than $W$, with the desired user having relatively small projection onto these modes. Thus, linear suppression of dominant interference modes can be accomplished with small noise enhancement. We show that similar observations apply for MIMO-OFDM over wideband multipath channels synthesized from measured 28 GHz data. We propose, and evaluate via information-theoretic benchmarks, per-subcarrier reduced dimension beamspace LMMSE in this setting.

This work proposes a unified control architecture that couples a Reinforcement Learning (RL)-driven controller with a disturbance-rejection Extended State Observer (ESO), complemented by an Event-Triggered Mechanism (ETM) to limit unnecessary computations. The ESO is utilized to estimate the system states and the lumped disturbance in real time, forming the foundation for effective disturbance compensation. To obtain near-optimal behavior without an accurate system description, a value-iteration-based Adaptive Dynamic Programming (ADP) method is adopted for policy approximation. The inclusion of the ETM ensures that parameter updates of the learning module are executed only when the state deviation surpasses a predefined bound, thereby preventing excessive learning activity and substantially reducing computational load. A Lyapunov-oriented analysis is used to characterize the stability properties of the resulting closed-loop system. Numerical experiments further confirm that the developed approach maintains strong control performance and disturbance tolerance, while achieving a significant reduction in sampling and processing effort compared with standard time-triggered ADP schemes.

Differentially flat models are frequently used to design feedforward controllers for electromechanical systems. However, control performance depends on model accuracy, which makes feedback imperative. This paper presents a control scheme for electromechanical systems in which measuring or estimating the output to be controlled – typically the position – is not feasible. It employs an identifiable-model-based controller and predictor, coupled with an iterative loop that updates model parameters using the error between a measurable output and its prediction. Simulations on electromechanical switching devices show effective tracking of the desired position trajectory using only coil current measurements.

Energy and pollution are urging problems of the 21th century. By gradually changing the actual power grid system, smart grid may evolve into different systems by means of size, elements and strategies, but its fundamental requirements and objectives will not change such as optimizing production, transmission, and consumption. Studying the smart grid through modeling and simulation provides us with valuable results which cannot be obtained in real world due to time and cost related constraints. Moreover, due to the complexity of the smart grid, achieving global optimization is not an easy task. In this paper, we propose a complex system based approach to the smart grid modeling, accentuating on the optimization by combining game theoretical and classical methods in different levels. Thanks to this combination, the optimization can be achieved with flexibility and scalability, while keeping its generality.

Objective: Surface electromyography (EMG) is a non-invasive sensing modality widely used in biomechanics, rehabilitation, prosthetic control, and human-machine interfaces. Despite decades of use, achieving robust generalization across subjects, recording systems, and acquisition protocols remains challenging. While foundation models (FMs) are gaining traction for EMG, existing approaches remain limited to single downstream tasks and lack deployability on embedded platforms. This work addresses these limitations. Methods: We present TinyMyo, a lightweight FM based on a Transformer encoder architecture. The model is pre-trained in a self-supervised manner using masked reconstruction on publicly available datasets. With only 3.6M parameters, TinyMyo is designed to support multiple downstream tasks through minimal task-specific head adaptations. Results: We demonstrate generalization across hand gesture classification, hand kinematic regression, speech production and speech recognition, with performance comparable to or surpassing the state of the art (SoA), and model size below 5M parameters. We achieve SoA results compared to previous FM-based works on the NinaPro DB5 (89.4%), UCI-EMG (97.56%), and EPN-612 (96.74%) datasets. We demonstrate the first-time deployment of an EMG FM on an ultra-low power microcontroller (GAP9), with an inference time of 0.785 s, energy of 44.91 mJ and power envelope of 57.18 mW. Conclusion: TinyMyo demonstrates that compact, self-supervised EMG FM can guarantee strong generalization across multiple downstream tasks while remaining compatible with low-power edge devices. Significance: TinyMyo is the first EMG FM for ultra-low power edge devices, enabling scalable and energy-efficient sensing for motor intent decoding, neuromuscular assessment, and biosignal driven human-machine interaction.

The use of synthetically generated data for training models is becoming a common practice. While generated data can augment the training data, repeated training on synthetic data raises concerns about distribution drift and degradation of performance due to contamination of the dataset. We investigate the consequences of this bootstrapping process through the lens of continual learning, drawing a connection to Generative Experience Replay (GER) methods. We present a statistical analysis showing that synthetic data introduces significant bias and variance into training objectives, weakening the reliability of maximum likelihood estimation. We provide empirical evidence showing that popular generative models collapse under repeated training with synthetic data. We quantify this degradation and show that state-of-the-art GER methods fail to maintain alignment in the latent space. Our findings raise critical concerns about the use of synthetic data in continual learning.

We introduce Guided Harmonic Path-Integral Diffusion (GH-PID), a linearly-solvable framework for guided Stochastic Optimal Transport (SOT) with a hard terminal distribution and soft, application-driven path costs. A low-dimensional guidance protocol shapes the trajectory ensemble while preserving analytic structure: the forward and backward Kolmogorov equations remain linear, the optimal score admits an explicit Green-function ratio, and Gaussian-Mixture Model (GMM) terminal laws yield closed-form expressions. This enables stable sampling and differentiable protocol learning under exact terminal matching. We develop guidance-centric diagnostics – path cost, centerline adherence, variance flow, and drift effort – that make GH-PID an interpretable variational ansatz for empirical SOT. Three navigation scenarios illustrated in 2D: (i) Case A: hand-crafted protocols revealing how geometry and stiffness shape lag, curvature effects, and mode evolution; (ii) Case B: single-task protocol learning, where a PWC centerline is optimized to minimize integrated cost; (iii) Case C: multi-expert fusion, in which a commander reconciles competing expert/teacher trajectories and terminal beliefs through an exact product-of-experts law and learns a consensus protocol. Across all settings, GH-PID generates geometry-aware, trust-aware trajectories that satisfy the prescribed terminal distribution while systematically reducing integrated cost.

Diffusion-based samplers – Score Based Diffusions, Bridge Diffusions and Path Integral Diffusions – match a target at terminal time, but the real leverage comes from choosing the schedule that governs the intermediate-time dynamics. We develop a path-wise schedule – selection gramework for Harmonic PID with a time-varying stiffness, exploiting Piece-Wise-Constant(PWC) parametrizations and a simple hierarchical refinement. We introduce schedule-sensitive Quality-of-Sampling (QoS) diagnostics. Assuming a Gaussian-Mixture (GM) target, we retain closed-form Green functions' ration and numerically stable, Neural-Network free oracles for predicted-state maps and score. Experiments in 2D show that QoS driven PWC schedules consistently improve early-exit fidelity, tail accuracy, conditioning of the dynamics, and speciation (label-selection) timing at fixed integration budgets.

This work introduces a single-range-aided navigation observer that reconstructs the full state of a rigid body using only an Inertial Measurement Unit (IMU), a body-frame vector measurement (e.g., magnetometer), and a distance measurement from a fixed anchor point. The design first formulates an extended linear time-varying (LTV) system to estimate body-frame position, body-frame velocity, and the gravity direction. The recovered gravity direction, combined with the body-frame vector measurement, is then used to reconstruct the full orientation on $\mathrm{SO}(3)$, resulting in a cascaded observer architecture. Almost Global Asymptotic Stability (AGAS) of the cascaded design is established under a uniform observability condition, ensuring robustness to sensor noise and trajectory variations. Simulation studies on three-dimensional trajectories demonstrate accurate estimation of position, velocity, and orientation, highlighting single-range aiding as a lightweight and effective modality for autonomous navigation.

In the process industries, MPC (Model Predictive Control) is typically implemented as a two-stage controller with a Linear Program (LP) steady-state optimizer that generates economically optimal targets for the MPC algorithm. Abnormal behaviors in industrial LP optimizers are often difficult to rationalize, especially when a large number of manipulated variables (MVs) and controlled variables (CVs) are involved. We introduce a novel, post-hoc LP explainability method by recasting the role of shadow prices in the LP solution as an attribution mechanism for MV-CV relationships. The core idea is that the shadow price of a constrained CV is not just an intrinsic property of the LP solution, but can be split into contributions from individual unconstrained MVs and resolved into one-to-one MV-CV pairings using a linear sum assignment algorithm. The proposed MV-CV pairing framework serves as a practical explainability tool for online LP-MPC systems, enabling practitioners to diagnose suboptimal constraints and verify alignment of the controller's behavior with its original design.

Food drying is widely used to reduce moisture content, ensure safety, and extend shelf life. Color evolution of food samples is an important indicator of product quality in food drying. Although existing studies have examined color changes under different drying conditions, current approaches primarily rely on low-dimensional color features and cannot fully capture the complex, dynamic color trajectories of food samples. Moreover, existing modeling approaches lack the ability to generalize to unseen process conditions. To address these limitations, we develop a novel multi-modal color-trajectory prediction method that integrates high-dimensional temporal color information with drying process parameters to enable accurate and data-efficient color trajectory prediction. Under unseen drying conditions, the model attains RMSEs of 2.12 for cookie drying and 1.29 for apple drying, reducing errors by over 90% compared with baseline models. These experimental results demonstrate the model's superior accuracy, robustness, and broad applicability.

The adversarial worst-case load shedding (AWLS) problem is pivotal for identifying critical contingencies under line outages. It is naturally cast as a bilevel program: the upper level simulates an attacker determining worst-case line failures, and the lower level corresponds to the defender's generator redispatch operations. Conventional techniques using optimality conditions render the bilevel, mixed-integer formulation computationally prohibitive due to the combinatorial number of topologies and the nonconvexity of AC power flow constraints. To address these challenges, we develop a novel single-level optimal value-function (OVF) reformulation and further leverage a data-driven neural network (NN) surrogate of the follower's optimal value. To ensure physical realizability, we embed the trained surrogate in a physics-constrained NN (PCNN) formulation that couples the OVF inequality with (relaxed) AC feasibility, yielding a mixed-integer convex model amenable to off-the-shelf solvers. To achieve scalability, we learn a sparse, area-partitioned NN via spectral clustering; the resulting block-sparse architecture scales essentially linearly with system size while preserving accuracy. Notably, our approach produces near-optimal worst-case failures and generalizes across loading conditions and unseen topologies, enabling rapid online recomputation. Numerical experiments on the IEEE 14- and 118-bus systems demonstrate the method's scalability and solution quality for large-scale contingency analysis, with an average optimality gap of 5.8% compared to conventional methods, while maintaining computation times under one minute.

Autonomous driving in interactive traffic scenarios remains challenging because of the mutual influence among vehicles and the inherent uncertainty of surrounding agents. Several model predictive control (MPC) formulations have been proposed to address this challenge, each adopting a different model of inter-agent interaction. While higher-fidelity interaction models enable more intelligent behavior, they incur substantially greater computational cost. Since strong interactions arise only occasionally in real traffic, a practical strategy for balancing performance and computational overhead is to invoke an appropriate controller based on situational demands. To this end, we first conduct a comparative study to assess and hierarchize the interactive capabilities of different MPC formulations. Building on this hierarchy, we then develop a neural network-based classifier for situation-aware switching among these controllers. We demonstrate that, by invoking the most advanced interactive MPC only in rare but critical situations and relying on a basic MPC in the majority of situations, situation-aware switching substantially improves overall performance while significantly reducing computational load.

This paper presents a real-time control framework for nonlinear pure-feedback systems with unknown dynamics to satisfy reach-avoid-stay tasks within a prescribed time in dynamic environments. To achieve this, we introduce a real-time spatiotemporal tube (STT) framework. An STT is defined as a time-varying ball in the state space whose center and radius adapt online using only real-time sensory input. A closed-form, approximation-free control law is then derived to constrain the system output within the STT, ensuring safety and task satisfaction. We provide formal guarantees for obstacle avoidance and on-time task completion. The effectiveness and scalability of the framework are demonstrated through simulations and hardware experiments on a mobile robot and an aerial vehicle, navigating in cluttered dynamic environments.

This paper addresses the problem of estimating air velocity and full attitude for unmanned aerial vehicles (UAVs) in GNSS-denied environments using minimal onboard sensing-an interesting and practically relevant challenge for UAV navigation. The contribution of the paper is twofold: (i) an observability analysis establishing the conditions for uniform observability, which are useful for trajectory planning and motion control of the UAV; and (ii) the design of a nonlinear observer on SO3R3R that incorporates pitot-tube, barometric altitude, and magnetometer measurements as outputs, with IMU data used as inputs, within a unified framework. Simulation results are presented to confirm the convergence and robustness of the proposed design, including under minimally excited trajectories.

Curve-straight probabilistic engagement zones (CSPEZ) quantify the spatial regions an evader should avoid to reduce capture risk from a turn-rate-limited pursuer following a curve-straight path with uncertain parameters including position, heading, velocity, range, and maximum turn rate. This paper presents methods for generating evader trajectories that minimize capture risk under such uncertainty. We first derive an analytic solution for the deterministic curve-straight basic engagement zone (CSBEZ), then extend this formulation to a probabilistic framework using four uncertainty-propagation approaches: Monte Carlo sampling, linearization, quadratic approximation, and neural-network regression. We evaluate the accuracy and computational cost of each approximation method and demonstrate how CSPEZ constraints can be integrated into a trajectory-optimization algorithm to produce safe paths that explicitly account for pursuer uncertainty.

This paper presents a geometric and theoretical study of an exponentially varying look-ahead parameter for UAV path-following guidance. Conventional guidance laws with a fixed look-ahead distance often drive the vehicle into turn-rate saturation when the heading or cross-track error is large, leading to constrained maneuvers and higher control effort. The proposed variable L0 strategy reshapes the look-ahead profile so that the guidance command adapts to the evolving tracking error geometry. A detailed investigation shows that this adaptation significantly enlarges the region in which the commanded turn rate remains unsaturated, allowing the vehicle to operate smoothly over a broader range of error conditions. For representative settings, the unsaturated operational envelope increases by more than 70% relative to the constant L0 formulation. These geometric insights translate to smoother trajectories, earlier recovery from saturation, and reduced control demand. Simulation studies on straight-line and elliptical paths demonstrate the merits of the variable look-ahead strategy, highlighting its control-efficient and reliable path-following performance.

This paper develops a unified perspective on several optimal control formulations through the lens of Kullback-Leibler (KL) regularization. We propose a central problem that separates the KL penalties on policies and transitions with independent weights, thus generalizing the standard trajectory-level KL-regularization used in probabilistic optimal control. This umbrella formulation recovers various control problems: the classical Stochastic Optimal Control (SOC), Risk-Sensitive Stochastic Optimal Control (RSOC), and their policy-based KL-regularized counterparts, termed soft-policy SOC and RSOC, which yield tractable surrogates. Beyond being regularized variants, these soft-policy formulations majorize the original SOC and RSOC, thus, iterating their solutions recovers the original objectives. We further identify a synchronized case of soft-policy RSOC where the policy and transition KL weights coincide, yielding a linear Bellman operator, path-integral solution, and compositionality – extending these computationally favourable properties to a broad class of control problems.

Wearable physiological signals exhibit strong nonlinear and subject-dependent behavior, challenging traditional linear models. This study provides a unified evaluation of cognitive load, stress, and physical exercise recognition using three public Empatica~E4 datasets. Across all conditions, nonlinear machine learning models consistently outperformed linear baselines, achieving 0.89–0.98 accuracy and 0.96–0.99 ROC–AUC, while linear models remained below 0.70–0.73 AUC. Although Leave-One-Subject-Out validation revealed substantial inter-individual variability, nonlinear models maintained moderate cross-person generalization. Ablation and statistical analyses confirmed the necessity of multimodal fusion, particularly EDA, temperature, and ACC, while SHAP interpretability validated these findings by uncovering physiologically meaningful feature contributions across tasks. Overall, the results demonstrate that physiological state recognition is fundamentally nonlinear and establish a unified benchmark to guide the development of more robust wearable health-monitoring systems.

This paper presents our work for the ICASSP 2026 Environmental Sound Deepfake Detection (ESDD) Challenge. The challenge is based on the large-scale EnvSDD dataset that consists of various synthetic environmental sounds. We focus on addressing the complexities of unseen generators and low-resource black-box scenarios by proposing an audio-text cross-attention model. Experiments with individual and combined text-audio models demonstrate competitive EER improvements over the challenge baseline (BEATs+AASIST model).

Current speech-language models (SLMs) typically use a cascade of speech encoder and large language model, treating speech understanding as a single black box. They analyze the content of speech well but reason weakly about other aspects, especially under sparse supervision. Thus, we argue for explicit reasoning over speech states and actions with modular and transparent decisions. Inspired by cognitive science we adopt a modular perspective and a world model view in which the system learns forward dynamics over latent states. We factorize speech understanding into four modules that communicate through a causal graph, establishing a cognitive state search space. Guided by posterior traces from this space, an instruction-tuned language model produces a concise causal analysis and a user-facing response, enabling counterfactual interventions and interpretability under partial supervision. We present a graph-based modular speech model for explicit reasoning, highlighting a path toward more transparent and controllable speech understanding.

Accurate real-time waypoints estimation for the UAV-based online Terrain Following during wildfire patrol missions is critical to ensuring flight safety and enabling wildfire detection. However, existing real-time filtering algorithms struggle to maintain accurate waypoints under measurement noise in nonlinear and time-varying systems, posing risks of flight instability and missed wildfire detections during UAV-based terrain following. To address this issue, a Residual Variance Matching Recursive Least Squares (RVM-RLS) filter, guided by a Residual Variance Matching Estimation (RVME) criterion, is proposed to adaptively estimate the real-time waypoints of nonlinear, time-varying UAV-based terrain following systems. The proposed method is validated using a UAV-based online terrain following system within a simulated terrain environment. Experimental results show that the RVM-RLS filter improves waypoints estimation accuracy by approximately 88$\%$ compared with benchmark algorithms across multiple evaluation metrics. These findings demonstrate both the methodological advances in real-time filtering and the practical potential of the RVM-RLS filter for UAV-based online wildfire patrol.

Deep residual networks (ResNets) have demonstrated outstanding success in computer vision tasks, attributed to their ability to maintain gradient flow through deep architectures. Simultaneously, controlling the Lipschitz constant in neural networks has emerged as an essential area of research to enhance adversarial robustness and network certifiability. This paper presents a rigorous approach to the general design of $\mathcal{L}$-Lipschitz deep residual networks using a Linear Matrix Inequality (LMI) framework. Initially, the ResNet architecture was reformulated as a cyclic tridiagonal LMI, and closed-form constraints on network parameters were derived to ensure $\mathcal{L}$-Lipschitz continuity; however, using a new $LDL^\top$ decomposition approach for certifying LMI feasibility, we extend the construction of $\mathcal{L}$-Lipchitz networks to any other nonlinear architecture. Our contributions include a provable parameterization methodology for constructing Lipschitz-constrained residual networks and other hierarchical architectures. Cholesky decomposition is also used for efficient parameterization. These findings enable robust network designs applicable to adversarial robustness, certified training, and control systems. The $LDL^\top$ formulation is shown to be a tight relaxation of the SDP-based network, maintaining full expressiveness and achieving 3%-13% accuracy gains over SLL Layers on 121 UCI data sets.

The Brunovsky canonical form provides sparse structural representations that are beneficial for computational optimal control, yet existing methods fail to compute it reliably. We propose a technique that produces Brunovsky transformations with substantially lower construction errors and improved conditioning. A controllable linear system is first reduced to the staircase form via an orthogonal similarity transformation. We then derive a simple linear parametrization of the transformations yielding the unique Brunovsky form. Numerical stability is further enhanced by applying a deadbeat gain before computing system matrix powers and by optimizing the linear parameters to minimize condition numbers.

Body-fixed spacecraft thrust couples attitude tracking error to translational motion because the applied inertial acceleration depends on spacecraft attitude. In this work, we formulate the dynamics of a continuously thrusting spacecraft on $\mathrm{SE}_2(3)$ to make this coupling explicit while retaining a structured form suitable for analysis. We represent attitude, velocity, and position within a single Lie-group state and derive the logarithmic dynamics of a left-invariant tracking error with position-dependent gravity. The resulting dynamics show explicitly how attitude error affects translational velocity through the applied acceleration and separate the structured linear dynamics from nonlinear contributions due to body-frame input mismatch and position-dependent gravity difference. We specialize the gravity contribution to the two-body model and derive an analytical bound in terms of the reference orbital radius, logarithmic position error, and attitude error. Numerical simulations of a continuous-thrust maneuver in a highly elliptical Molniya-like orbit show agreement between the derived logarithmic error dynamics and errors reconstructed from independent nonlinear state propagation.

The optimal operation of modern microgrids, particularly those integrating stochastic renewable generation and battery energy storage system (BESS), relies heavily on load and disturbances forecasting to minimize operational costs. However, in environments with uncertainties in both generation and consumption, traditional numerical forecasting methods often fail to capture generation shifts and event-driven load surges. While contextual information regarding event schedules, system logs, and computational task records is easily obtainable, classic control paradigms lack a formal interface to integrate the unstructured, semantic data into the physical operation loop. This paper addresses this gap by introducing the InstructMPC framework, which utilizes a Large Language Model (LLM) paired with a tunable last layer mapping to translate unstructured operational context into predictive disturbance trajectories for the MPC controller. Unlike conventional forecasting methods, the proposed approach treats the last layer mapping as a tunable component, refined online based on the realized control cost. We establish a theoretical foundation for this closed-loop tuning strategy, proving a regret bound of $O(\sqrt{T \log T})$ for linear systems under a tailored task-aware loss function, together with robustness guarantees against uninformative or noisy textual inputs. The control strategy is experimentally validated on OpenCEM, a real-world microgrid with highly fluctuating generation and consumption. Experimental results demonstrate that the LLM-driven MPC significantly reduces cumulative grid electricity costs compared to classical context-agnostic baselines, validating the efficacy of integrating semantic information directly into physical control loops.