Papers

In offline reinforcement learning (RL), the distribution shift between behavioral data and the learned policy can lead to erroneous Q-value estimation, thereby misguiding the direction of policy optimization. To address this issue, we develop a behavioral advantage corrected policy evaluation (BAC-PE) approach, which utilizes the Q-function of the behavior policy to correct the learned policy's Q-function, thus mitigating pessimistic conservatism and overestimation bias. Furthermore, the convergence of BAC-PE is analyzed theoretically, and an upper bound on the difference between the learned Q-function and the true Q-function is derived. To alleviate distribution shift, this work employs diffusion models to represent both the behavior policy and the learned policy, performing distribution matching for accurate policy regularization. Additionally, Q-value guidance is incorporated into the training process to achieve effective policy improvement. By combining BAC-PE with diffusion policy modeling, we propose the diffusion policy with behavioral advantage correction (DPBAC) algorithm. Compared to existing offline methods, DPBAC demonstrates stronger policy representation capabilities and effectively mitigates the bias in Q-value estimation. Experimental results on multiple domains of D4RL tasks show that DPBAC achieves superior performance, with notable advantages over state-of-the-art (SOTA) algorithms.

We present a set-based framework for estimating human cognitive states in human-automation interaction (HAI) contexts. Unlike probabilistic approaches dominant in the HAI literature, our framework treats process and measurement uncertainties as unknown but bounded, avoiding the need for large structured datasets or distributional assumptions on noise. We demonstrate the framework in the context of conditionally automated (SAE Level 3) driving, where we estimate three cognitive states (trust, perceived risk, and workload) that influence human reliance on the automation during a continuous, non-trial-based interaction. We leverage a hybrid dynamical modeling framework to identify individual-specific process and measurement models, systematically estimate noise bounds by enforcing reachset conformance, and identify the subset of cognitive states that influence each individual's reliance on the automation. Set-valued estimates of those states are then produced by fusing binary reliance observations and intermittent, quantized self-reports. The framework is evaluated through an in-person experiment in a medium-fidelity driving simulator with 20 participants. The set-valued estimator achieves at least 75% consistency for most participants during testing, and multi-step-ahead reliance predictions derived from the estimates outperform both an open-loop baseline and a particle filter across all choices of prediction horizons (15, 30, 45, 60 time steps), with the performance gap more evident at longer horizons. The proposed estimation framework can enable automation systems that are continuously aware of, and responsive to, the human driver's state.

This paper studies a resource-allocation inefficiency in batched large language model (LLM) serving: heterogeneous requests that share a decode batch impose max-driven computational costs on one another. Because the wall-clock cost of a batch step is largely governed by the largest active KV-cache footprint, a short request co-batched with a long request can experience latency and GPU-resource consumption disproportionate to its own token workload. We formalize this phenomenon as a resource-fair scheduling problem. We develop a mathematical scheduling model that connects within-batch resource fairness to system throughput. The proposed fairness constraint bounds the disparity in decode progress, equivalently KV-cache footprint, among co-batched requests. Based on this model, we design the Insert-Short-Jobs-with-Limit (ISJL) algorithm, a parameterized hybrid batching policy. We prove that ISJL achieves a global competitive-ratio lower bound of $3/4$. We further examine the profit implications of resource-fair scheduling under the token-metered pricing convention used by commercial LLM APIs. Numerical experiments show that ISJL occupies a favorable middle ground between FCFS, which has large batching externalities, and LJF, which is cost-aligned but sacrifices batching flexibility. Thus, ISJL provides a bi-criterion scheduling policy: it maintains high throughput while aligning max-driven batch cost with token-metered revenue.

Reliable photovoltaic (PV) forecasts can support low-carbon energy systems, but deployed sites may have only short and incomplete records. Physical and hybrid methods can be sensitive to weather inputs, calibration, and timestamp-alignment, while individual machine learning models may capture different parts of the forecasting problem. We study hourly day-ahead PV forecasting at a United Kingdom charging station using one year of inverter measurements, with 9.25% of hours missing. The pipeline checks timestamp-alignment, derives solar and clearness features, adds short-term weather context, and combines five complementary models using non-negative least squares stacking, with the combination fitted only on validation observations. We compare against smart persistence, a weather-scaled baseline that carries the previous day's PV behaviour forward using target-day irradiance. With retrospective weather, the combined model reduces daylight normalised root mean square error (RMSE) by 31.2% under random day-fold evaluation and by 2.9% under rolling-origin evaluation, although the latter improvement is not robust across days. It also improves by 3.0% over the single model selected from validation performance. Replacing retrospective weather with a public product sampled at a constant 24-hour lead increases daylight RMSE by 13.1% and 4.2% under the two protocols, while retaining positive skill over smart persistence.

In vision-aided navigation and visual-inertial odometry, the quality of triangulated three-dimensional feature positions is a fundamental prerequisite for state estimation accuracy. Triangulation becomes ill-conditioned or even impossible when a camera undergoes pure rotation without translation, or when the observed bearing vectors provide insufficient parallax. Even though visual-inertial odometry has been extensively studied, the active maintenance of feature-observability during navigation has not been sufficiently addressed in the literature. To address this gap, this study presents TANGO-VIO, a triangulation-aware navigation framework that embeds a log-determinant metric of the feature-wise stacked-bearing matrix into a control barrier function. In this proposed method, the observability guarantee is established in the feature-geometric sense by enforcing a lower bound on the aggregate triangulation-information metric through a nominal-direction-weighted minimum-deviation velocity correction. The proposed architecture is evaluated through software-inthe- loop simulations and real flight experiments. The results show improved triangulation conditioning under low-parallax motion, while the flight response closely reproduces the corresponding simulation behavior and confirms the practical realizability of the proposed safety filter. Supplementary materials are available on the project webpage.

Developing deployable locomotion policies through conventional reinforcement learning often requires complex reward engineering and expensive training times. While differentiable simulation offers a highly efficient alternative, open-source tools capable of end-to-end transfer of these policies to physical hardware remain limited. This paper introduces Open-DiffLoco, an open-source framework for training deployable blind quadruped locomotion policies with differentiable simulation. The framework implements the Short-Horizon Actor-Critic (SHAC) algorithm in MuJoCo XLA (MJX) and trains a proprioceptive policy that transfers to real-world hardware. The deployed policy removes privileged actor observations, including base linear velocity, and does not rely on reference trajectories. It also uses a substantially simplified reward function, enabling the robot to discover walking patterns without the complex auxiliary rewards typically used in conventional reinforcement learning pipelines. When deployed on physical hardware (a Unitree Go2 quadruped), the trained policy tracks omnidirectional velocity commands with root-mean-square error below 0.2 m/s, reaches speeds above 1 m/s, and remains robust to uneven terrain and external physical disturbances, such as lateral pushes. Across the reported configurations, training uses under 6 GB of VRAM on a single NVIDIA GeForce RTX 5080 GPU and completes in approximately 20-60 minutes. As an algorithmic extension to SHAC, we propose Jacobian-Augmented Value Estimation (JAVE), which supervises the critic Jacobians to improve early first-order policy-gradient training. To our knowledge, Open-DiffLoco is the first open-source framework for training deployable locomotion policies using differentiable simulation. Deployment videos and source code are available at: https://diffloco.martin-opat.com/

Cervantes-Pérez et al. (arXiv:2601.12545) claim, on experimental grounds, that high-gain injection is practically inadmissible, taking the SM observer of Davila, Fridman & Poznyak as a prototypical example. We re-simulate their plant, trajectory and gains with explicit Euler integration at \Delta t = 1 ms and 0.1 ms, and implement both observers on independent hardware. We got three findings: i) the controller gains k_p=1600, k_v=1100 place the tracking-error poles at the "pathological" locations -1.455 and -1098.5 (\zeta = 13.7); the super-twisting scheme lands on the resulting nominal response and is invariant to its own observer gain over a fifty-fold range; ii) the I&I margin is monotone in k_v and reverses near k_v \approx 170, where the loop pole crosses the induced observer pole at -9.0 rad/s; iii) refining the sampling step tenfold moves every I&I figure by a factor of three, while the SM figures move by 0.2%. On hardware, a signum-based super-twisting observer tuned at the Levant-Moreno perturbation bound outperforms I&I on ISE, ITAE and ISC simultaneously, without audible chattering, on a $60 motor with a 0.043^\circ encoder at 500 Hz. The gains of Cervantes-Perez et al. exceed the same prescription by two orders of magnitude; the chattering the authors report is the designed consequence of that excess.

State-dependent switching diffusions can be contractive in distribution even when some modes are individually expansive. We develop a computable transport-based condition for such contraction on $\mathbb{R}^n$ and on compact convex domains with normal reflection. The transport cost combines mode-dependent spatial weights with a discrete mode penalty while preserving spatial separation for cross-mode pairs. Using synchronous coupling of the Brownian motions and maximal coupling of the state-dependent jump clocks, we derive separate generator inequalities for same-mode and cross-mode configurations. Convex normal reflection contributes a nonpositive finite-variation term, so the same conditions apply to the associated no-flux Fokker-Planck-Kolmogorov system. Their feasibility guarantees global pairwise exponential contraction of the Markov semigroup and weak measure solutions, with unit prefactor and an explicit rate. For a fixed ordering of the mode weights and a prescribed decay rate, the conditions are affine in the spatial weights and graph costs and form a semi-infinite linear feasibility problem. A finite-mesh condition with a Lipschitz margin certifies the inequalities over the full domain. A reflected one-dimensional example validates the distributional computation, and a planar three-mode example demonstrates the synthesis procedure for transition rates depending on both state coordinates.

This work presents a human-centered collaborative framework that integrates Preference-Based Optimization (PBO) and Dynamic Movement Primitives (DMPs) to optimize robot-assisted tasks such as painting. The system allows the operator to perform the process while the robot adapts its behavior in real-time, dynamically adjusting the orientation of the piece in order to match the orientation of the operator's hand. The PBO framework leverages the GLISp algorithm to iteratively refine control parameters such as execution time, robot responsiveness, and rotation amplification through human feedback. Moreover, DMPs have been modified to enhance the reactive behavior of the robot and its adaptability to ergonomic requirements. The method was validated with a heterogeneous group of participants executing \rev{painting tasks}. The results show that our strategy effectively reduces operator effort while optimizing process outcomes.

Passive overlay communication for batteryless devices is an important enabling capability for next-generation vehicle-to-everything (V2X) networks. However, enabling reliable passive payload delivery without occupying additional spectrum remains challenging, since overlay signaling must be embedded into short and time-varying vehicular packets while preserving the decodability of the legacy host transmission. This paper investigates a packetized batteryless V2X overlay architecture in which a dedicated short-range communications (DSRC)-based packet simultaneously carries conventional V2X data and a passive overlay payload. A compact PHY-layer model is developed to characterize the coupled effects of attenuation depth, embedded-bit rate, and legacy modulation and coding scheme (MCS) on host-link and passive-link reliability, as well as packet-level embedding feasibility. We then formulate a sum-throughput maximization problem that jointly accounts for the legacy packet error rate and passive decoding error rate. We further propose a multi-agent reinforcement learning (MARL)-based adaptive parameter-selection method. Simulation results show that the proposed MARL controller achieves stable convergence and improves the average throughput by 15%, demonstrating the effectiveness of throughput-driven PHY adaptation for batteryless V2X overlay communications.

Fast dynamic obstacle avoidance (DOA) on uncrewed aerial vehicles (UAVs) demands not only low-latency control and actuation but also reliable perception with sufficient sensing range for accurate obstacle detection and speed estimation. This letter presents, to the best of our knowledge, the first mmWave RADAR-based perception-and-control system for fast onboard DOA. We derive and analyze latency and spatial bounds that relate sensing range, relative speed, and control delay, yielding sufficient conditions for successful avoidance. Our system adopts a lightweight tracker based on interacting multiple models and a controller based on control-barrier functions that directly outputs evasive accelerations. It achieves position errors of less than 0.15 m, 0.93 m, and 0.87 m in x, y, and z directions for 300 experiments with three different object sizes and varying visibility (light and dark), and a similar spread for 90 experiments in smoke. An onboard implementation on a Raspberry Pi 4B demonstrates real-time feasibility with an end-to-end sensing-to-command latency of approximately 14 ms. Code and the full dataset of 390 throws are available (https://tinyurl.com/radardoagit).

To improve the operational readiness of combat aircraft engines and reduce unplanned maintenance costs, accurately estimating the remaining useful life (RUL) is critical. Traditional maintenance often proves insufficient under dynamic mission profiles. In this study, a deep learning-based predictive maintenance model capable of autonomously extracting features from multivariate sensor data was developed. Using the NASA C-MAPSS FD001 and FD004 datasets, data were converted into sequential blocks via 50- and 30-step sliding windows, respectively. The model's architectural superiority in autonomously extracting temporal degradation features was validated against RF, CNN-LSTM, and BiLSTM baselines. On FD001, it achieved an R-squared (R2) of 0.8901, a 13.28 RMSE, and a 320.34 NASA risk score, demonstrating generalizability on the multi-regime FD004 dataset with a 15.71 RMSE. The proposed maintenance protocol achieved a 0.9973 AUC at the critical 30-cycle threshold, ensuring high reliability. Additionally, a decision-support simulator has been developed to validate this protocol under aggressive combat flight profiles.

Multi-link aerial robots can actively deform their articulated structures during flight, giving them strong potential for aerial manipulation. However, they still face substantial challenges in contact-rich aerial manipulation tasks such as surface sliding, which requires both disturbance robustness and compliance to uncertain surface geometry. Force-control strategies such as impedance and admittance control are commonly employed to address these requirements. Although impedance control can provide disturbance-resistant interaction and admittance control can offer compliant adaptation, their opposite force–motion causalities prevent their simultaneous implementation when applied through the same actuation source, such as the rotor thrusts used by conventional aerial robots. To overcome this limitation, we propose a hybrid impedance–admittance control strategy for a multi-link aerial robot. The articulated morphology enables a functional separation of force and motion regulation across joint and rotor actuation sources. In this framework, admittance behavior is generated through joint angle regulation to enhance adaptive interaction, while impedance behavior is achieved by modulating rotor thrust to regulate the sliding motion. This structural coordination allows the robot to leverage the complementary strengths of both control paradigms. As a result, the multi-link aerial robot achieves resilient and adaptive surface sliding. Experimental results demonstrate robust and compliant sliding performance on unknown surfaces.

In this paper we study feedback interconnections of positive nonlinear SISO systems and introduce two complementary properties: persistent-input/persistent-output (PIPO) for the plant and persistent-input/transient-output (PITO) for the controller. Assuming forward completeness of the closed loop and a one-sided affine growth bound on the controller output, we show that PIPO and PITO jointly imply boundedness of the control signal. The result is input-output in nature and does not require linearity or monotonicity of the interconnected subsystems, although it does not in general guarantee boundedness of the full controller state. We provide a structural PIPO condition for positive monotone plants with a class-\(\mathcal K_\infty\) steady-state characteristic, establish PITO for the antithetic integral controller with explicit gains, and illustrate the framework on a nonlinear integral-feedback motif with multiplicative controller growth.

This paper addresses the sim-to-real gap in dynamics arising when a parallel-link mechanism is represented by a serial-tree surrogate in simulation. Conventional Jacobian-based state and torque mappings preserve consistency with the kinematic and virtual-work relations but do not account for the coordinate-induced redistribution of actuator inertia and damping and the linkage inertia omitted during serial-tree reduction. To address this gap, Simulator-Side System Normalization (S3N) is proposed to normalize the serial-tree simulator's effective dynamics while preserving its tree topology. S3N-Act incorporates actuator inertia and damping into the serial-coordinate dynamics through coordinate transformation, whereas S3N-Full restores residual linkage inertia by separately identifying actuator- and leg-level frequency responses. In the 2-DoF validation, S3N-Full reduced the joint-position and torque RMSEs by 80.9% and 82.1%, respectively, relative to the Jacobian-mapping baseline. During pitch-in-place motion, S3N-Act and S3N-Full reduced the RMSE of the ground reaction force norm by 65.1% and 62.4%, respectively. During circular locomotion, S3N-Full reduced the phase-averaged, command-normalized sim-to-real gap from 17.3% to 9.9%. These results show that simulator-side normalization improves motion- and force-level sim-to-real consistency. It enables policy training in a serial-tree framework with hardware-consistent dynamics that better represent the physical parallel-link mechanism.

The typical common-bus structure and its inherent bus stabilization requirement can become a bottleneck for flexibility in power supplies for battery-powered autonomous systems, limiting scalability and plug-and-play capability. Power packetization overcomes this by eliminating the common bus through time-division multiplexing of physically isolated power flows. However, conventional power packet routers face two physical-layer limitations: uncontrollable transmission direction and limited controllability of the amount of energy per packet. This paper proposes a fully controllable bidirectional power packet transmission to address these limitations. We introduce an inter-router circuit featuring a parallel inductor as a temporary energy buffer, enabling power transmission in any direction with full control over the transferred energy. We derive the switching algorithm for the routers to perform this operation and verify its feasibility through experiments using prototype hardware.