Institute for Advanced Study

Grid-following (GFL) inverters face both severe largesignal destabilization (LSD) and small-signal destabilization (SSD) risks in ultra-weak grids. To address the LSD risk, this paper reveals the effect of reactive current injection (RCI) on the powerangle property, terminal voltage, and maximum transferable active power (MTAP) in the presence of different short-circuit ratio (SCR) conditions. Then, the feasible ranges for active power and RCI are developed to prevent LSD and terminal voltage violation. By excluding the possibility of LSD with an enhanced MTAP, the small-signal stability of the GFL inverter is ensured by applying the impedance stability criteria. In this article, an optimization problem with constraints of internal and external stabilities is established to redesign the gains of the current controller, without requiring real-time detection during the solution process. The redesigned gains effectively ensure the smallsignal stability of the GFL inverter while keeping a relatively high PLL bandwidth in an ultra-weak grid with SCR = 1.1, which is lower than that achieved by state-of-the-art methods. Finally, the simulation and experimental results confirm the effectiveness of the proposed guideline.

Fluorescence imaging (FLI) has become a cornerstone for biomedical research owing to its non-invasive, high sensitivity, and exceptional spatial resolution. The effectiveness of FLI is primarily governed by the properties of fluorescent probes, including photophysical characteristics, targeting capabilities, and responsiveness. However, current probe development in FLI relies on empirical trial-and-error strategies or quantum-chemical calculations, both of which are time-consuming, labor-intensive, and often insufficient for precisely deciphering complex structure-property-function relationships. Recent advances in artificial intelligence (AI) have opened a new avenue for rational probe design through rapid property prediction, high-throughput molecular screening, and performance-guided inverse design. Nevertheless, a systematic and comprehensive review elucidating AI-guided design of fluorescent probes in biomedical applications is still lacking. This review focuses on recent advances in AI-guided design and performance optimization of fluorescent probes for bioimaging applications. First, the conceptual workflow underlying AI-based predictive frameworks is elucidated. Second, the impact of AI on optimizing key probe properties-including optical performance, targeting, and responsive capabilities-is thoroughly elaborated. Third, the biomedical applications of AI-guided probes in imaging, sensing, and therapy are reviewed. Finally, the current challenges and future perspectives to accelerate the development of AI-guided fluorescent probe design are briefly discussed.

Accurate quantification of drug concentration within the skin's interstitial fluid (ISF) remains a significant analytical challenge due to the limitations of invasive sampling and the inability of bulk measurements to resolve micro-scale distribution. Traditionally, predictive models have treated the skin as a static barrier, ignoring the dynamic matrix effects caused by ISF flow, which leads to substantial errors in estimating deep-tissue analyte concentrations. To address this, this study proposes a computational analytical strategy integrating Finite Element Method (FEM) with Computational Fluid Dynamics (CFD) to quantitatively profile drug transport under varying thermal conditions. By calibrating against HPLC-validated ex vivo permeation data at a reference temperature, diffusion coefficients and ISF flow velocities were extrapolated to predict behavior at other temperatures. This approach effectively decouples the influence of fluid dynamics from passive diffusion, allowing for the precise resolution of temperature-dependent permeation kinetics. The Flow-Field model demonstrated strong correlations with ex vivo skin permeation tests, achieving R 2 values over 0.99 for various drugs and temperature conditions. This work establishes a robust in silico tool for the micro-scale profiling of analytes in complex biological tissues, offering a non-invasive alternative to estimate ISF concentrations where physical sampling is restricted. • Integrated strategy quantifies drug concentration in skin interstitial fluid. • FEM-CFD model decouples fluid dynamics for precise permeation analysis. • Flow-Field model achieves R 2 > 0.99 correlation with ex vivo skin permeation tests. • Temperature-dependent transdermal drug transport accurately predicted. • Non-invasive in silico tool resolves micro-scale analyte distribution.

Prosthetics, exoskeletons, and rehabilitation devices that seamlessly respond to the user's motion intent are essential for improving the quality of life for individuals with physical disabilities. However, existing motion intent recognition systems based on Electromyography (EMG) often experience significant performance degradation over time and limb positions. To address this, this paper proposed a Hierarchical EMG-FMG Fusion (HEFF) framework that incorporates Force Myography (FMG) as a complementary modality to enhance robustness in long-term and multi-position motion intent recognition. The HEFF framework introduced three new strategies: (1) Complementary Pyramid Fusion, a neural network architecture that effectively integrates the complementary characteristics of EMG and FMG; (2) Position Compensation and Pairing, to increase the generalizability of the model by producing augmented training data mimicking the temporal and positional variability; and (3) Feedback-based Baseline Correction, to dynamically counteract FMG baseline drift. The robustness of the system was evaluated through 10 distinct motions performed in 3 different limb positions, retested after an interval of approximately 7 days. Notably, the training data were only collected in a single limb position on the first day. The real-time experiments with 15 subjects showed that the HEFF framework maintained 91.89% classification accuracy in the primary limb position and 85.39% across multiple unseen positions. According to offline analysis, HEFF improved classification accuracy by up to 34.99% across various limb positions compared to the conventional fusion method. These results highlight HEFF's robustness across various conditions, being well-suited for assistive technology applications.

2D topological materials are garnering significant interest for their potential in optoelectronic applications, particularly due to their unique quantum transport properties and exceptional optical characteristics. In this work, quaternary topological BiSbSe2Te with stoichiometry modulation from Bi2Se3 are synthesized and demonstrates a broad spectral band photoresponse, ranging from infrared to terahertz and to millimeter waves, with a particular excellence on detection of low quantum energy photons. The observed photoresponse is attributed to the excitation of plasmonic nonequilibrium electrons originating from the topological Dirac surface states inherent in the BiSbSe2Te. The detector exhibits remarkable performance, achieving a responsivity of 8174V/W and a noise equivalent power of 4.7 × 10- 1 3 W/Hz1/2 at 0.094 THz, with a wide low quantum energy terahertz and millimeter wave photoresponse spanning from 0.032 to 0.173 THz. The study underscores the potential of Dirac surface-state driven photoresponse in BiSbSe2Te, paving the way for the development of sensitive, broad spectral band and room-temperature low quantum energy photodetectors, which are highly pursued for a variety of applications.

Inspired by the environment-adaptive behaviors of water striders, we 3D-printed a light-driven liquid crystal elastomer (LCE) swimming robot, OptiLCE Strider, capable of multimodal locomotion and adaptive reconfiguration at the air-water interface. Utilizing carbon nanotubes (CNTs) as photothermal fillers and dynamic disulfide bonds for shape reconfigurability, the robot exhibits three distinct propulsion modes: Marangoni-effect-driven continuous motion under low light intensity (1.3-7.2 mm s- 1), steam-wave-induced pulsatile locomotion under high light intensity (12.5-16.8mms- 1), and flapping propulsion enabled by reversible LCE deformation (4.6-6.9 mm s- 1). The dynamic disulfide bonds enable exceptional structural reconfigurability and environmental adaptability for the LCE robot to execute complex tasks, including maze navigation, cargo capture/transport, programmable rotation, and light-powered jumping (escape from grounded or obstructed states via actuation energy storage/release, with jumping height/distance 6×/3.3× the robot length). The qualitative phase map guides locomotion mode selection, while energetic cost analysis reveals a clear force-efficiency trade off among the three modes, guiding application specific selection. This study highlights the potential of dynamic LCE-based robots for intelligent systems in liquid interface environments, paving the way for versatile applications in soft robotics and biomimetic engineering.