This paper introduces K-MADDPG, a K-means enhanced Multi-Agent Deep Deterministic Policy Gradient algorithm, for solving static coverage tasks in drone swarms. The proposed approach implements a hierarchical control architecture that integrates K-means clustering for global task allocation with an enhanced MADDPG for local coordination, creating a two-layer control mechanism that optimizes both global coverage efficiency and local collaborative behavior. First, K-MADDPG employs K-means clustering to dynamically partition the target area into sub-regions, with each drone assigned to a cluster center via the Hungarian algorithm. This global allocation strategy ensures balanced coverage distribution and provides clear navigation objectives for each agent. Then, an enhanced MADDPG algorithm processes grid information to enable precise local decision-making. Simulation results demonstrate that KMADDPG outperforms traditional MADDPG in both convergence speed and coverage stability. The hierarchical architecture enables efficient adaptation to varying environment sizes, while the grid processing enhances spatial awareness and navigation precision. This approach effectively addresses the scalability and efficiency challenges faced by swarm robots in static coverage scenarios.

This paper presents a novel safety motion planning framework for mobile robots operating in dynamic environments, addressing the challenges of dynamic feasibility, safety robustness, and real-time responsiveness. The proposed framework integrates homotopy topology-based hierarchical search, FRS/BRS spatiotemporal corridor pruning, and FRS-based safety validation. By utilizing homotopy topology, the framework ensures global path optimization while reducing search redundancy. The FRS/BRS corridor pruning technique improves real-time performance by discarding infeasible or unsafe regions, achieving planning frequencies of 10-15 Hz. The framework was experimentally validated and compared against several state-of-the-art algorithms. The results demonstrate that the proposed framework outperforms the baseline algorithms with a 97.8 % success rate and a low collision rate of 2.2%, providing superior path quality and real-time performance in dynamic environments. The framework offers a promising solution for real-time, high-quality motion planning in dynamic indoor environments.

The localization of unmanned systems in degraded visual environments (DVEs) has emerged as a critical research challenge in recent years. This paper studies the issues of robust multi-modal fusion and sensor degradation for RGB-Thermal-Inertial Odometry frameworks. A dual-track front-end method combining deep-learning descriptors and KLT optical flow is designed, ensuring reliable cross-modal association and efficient temporal tracking. Meanwhile, a scene-aware feature adapter and an asynchronous factor graph are developed to address sensor anomalies through dynamic sensor balancing and seamless monocular-binocular transitions. The effectiveness and superiority of the proposed framework are validated through real-world experiments in lighting transitions and smoke scenarios, significantly outperforming state-of-the-art systems with high localization accuracy at real-time frequencies.

Overcoming the impacts of model uncertainties and external disturbances is a common challenge in fixed-wing UAV control system design. This paper introduces a novel fixed-wing UAV attitude control framework that integrates Barrier Function Super-Twisting Control (BFSTC) with Single-loop Incremental Nonlinear Dynamic Inversion (SINDI). By leveraging SINDI to linearize second-order attitude dynamics, we apply a BFSTC strategy featuring an adaptive gain that dynamically responds to disturbance magnitudes. This integrated approach exhibits heightened robustness against internal and external disturbances compared to the standalone INDI or Sliding Mode Control (SMC) schemes, while simultaneously achieving faster response and reducing the parameter tuning burden. Additionally, we provide a rigorous stability analysis under the saturated control gain and offer systematic guidelines for barrier function parameter selection. Comprehensive validation through numerical simulations and real-world flight experiments confirms that the proposed method delivers exceptional tracking precision, robustness and adaptability across diverse flight paths and disturbance scenarios.

MOTIVATION:DNA storage offers exceptional information density and archival longevity but is constrained by complex biochemical noise inherent to synthesis, storage, and sequencing. Conventional hard-decision error-correction schemes often rely on excessive redundancy to mitigate these imperfections, which significantly compromises storage efficiency and density. RESULTS:We present Polus, a Transformer-based enhancement framework that improves digital reliability through soft-decision decoding (SDD) without requiring encoder modification. At its core is SeqFormer, a Transformer-based channel model that synergizes sequence context with quality signals to generate calibrated per-base confidence scores, effectively transforming uncertain biochemical noise into informative "soft" erasures. In in silico benchmarks, Polus significantly upgrades mainstream DNA storage codecs. It reduces the sequencing coverage required for DNA Fountain by 38.9%-increasing effective physical density by approximately 80%-and eliminates persistent indel-induced errors in the Yin-Yang codec. Furthermore, it enables a targeted resequencing strategy that achieves full recovery with 99.9% less overhead than uniform deepening. Moreover, a nine-metric evaluation suite was employed to provide multi-dimensional quantitative comparisons of DNA storage codecs across reliability, density, and cost. Collectively, Polus provides a reproducible framework for context-aware decoding and system design guidance for DNA storage. AVAILABILITY AND IMPLEMENTATION:All source code of the Polus, including the SeqFormer implementation, codec algorithms, test data used, and the simulation pipeline is available on GitHub (https://github.com/dinglulu/Polus) and Zenodo (https://zenodo.org/communities/bioinfoszu/). A web hosted instance of Polus is available at https://polus.bioailab.net/polls/home. The SeqFormer model is also released as a standalone repository at https://github.com/dinglulu/SeqFormer and https://zenodo.org/communities/bioinfoszu/.

Structural variants (SVs) are a major source of genomic diversity, yet their discovery remains challenging due to repetitive genomic contexts, alignment ambiguity, and the trade-off between sequencing cost and read length. Here we introduce HitSV, which substantially improves SV discovery by implementing repetitiveness and signature density aware breakpoint recognition coupled with precise haplotype-resolved local assembly, thereby enabling base-resolution SV reconstruction and genotyping across various sequencing technologies. HitSV is 12-68% (long-read), 3%-36% (short-read) and 13% (hybrid-sequencing), respectively, more accurate than state-of-the-art SV callers across different coverages. Applying HitSV to the 1KGP Phase 4 cohort, we identified 31.5% more SVs, substantially reshaping allele-frequency landscapes. Notably, analysis of a large Chinese long-read cohort uncovers tandem repeat–mobile element composite arrays as a prevalent and multi-allelic class of complex SVs, highlighting composite repeat architectures as a fundamental hallmark of human genomes.

The low-altitude economy (LAE) has become a key application field for unmanned aerial vehicles (UAVs), where cooperative target localization is crucial for logistics, infrastructure inspection and emergency response. However, UAV-based localization in LAE faces dual challenges: unknown or maneuvering target motions, and non-Gaussian unknown-but-bounded (UBB) noise from electromagnetic interference and complex airflows. To address these problems, a linear interacting multi-model-set membership filter (IMM-SMF) is proposed for multi-UAV cooperative target localization. This method integrates the IMM framework for adaptive tracking of maneuvering targets, adopts SMF to robustly handle UBB noise, and uses a sequential fusion strategy to fuse multi-UAV observation data. A simple linear model ensures computational efficiency, which is vital for real-time LAE applications. Theoretical analysis verifies the method's strong robustness to UBB noise in maneuvering target tracking, and comparative experiments confirm its superiority in localization accuracy and stability over traditional single-model or single-UAV methods. This work provides a reliable and efficient localization solution for UAV cooperative systems in the LAE.

Although Li-rich layered oxides (LLOs) are widely favored due to their high capacity derived from oxygen anionic redox, the rapid degradation of the initial lattice shortens their cycle-life. This is mainly attributed to internal strain and irreversible degradation of the oxygen redox environment, leading to layered structure damage, phase transformation and oxygen vacancies, which accumulate rapidly with cycling. Mechanism exploration and modification development targeting individual factors have established the principles and effectiveness of transition metal (TM) dopes in the Li layer and cation vacancies in the TM layer for inhibiting layered phase degradation and promoting anionic reversible reactions, thereby inspiring coupling defects engineering. Here, we show that more permanent cycle-life (85.77 % capacity retention and 0.38 mV/cycle voltage decay after 500 cycles at 1C, 1C = 250 mAh g-1) can be achieved by constructing interlayer TM-vacancy coupling defects. The simultaneously obtained strong interlayer TM-O-TM ribbon, TM doping and TM-O interaction collectively and effectively maintain the layered framework and occupancy sequence, providing a stable coordination environment for oxygen redox. This work demonstrates the feasibility of constructing interlayer TM-vacancy coupling defects to pursue Li-rich cathodes with both high energy density and long cycle-life.

DNA storage offers exceptional information density and archival longevity, but is constrained by the complex, heterogeneous errors inherent to synthesis, storage, and sequencing. Conventional error-correction schemes often rely on excessive logical redundancy to mitigate these biochemical imperfections, thereby compromising storage efficiency. Here, we introduce Polus, a deep-learning-enabled platform that bridges the gap between biochemical constraints and digital reliability through soft-decision decoding. At its core is SeqFormer, a Transformer-based channel model that synergizes sequence context with quality signals to characterize platform-specific error profiles, generating calibrated per-base confidence scores. This mechanism transforms uncertain biochemical noise into informative “soft” erasures. In in silico benchmarks, Polus significantly enhances mainstream codecs: it reduces the sequencing coverage required for DNA Fountain by 38.9% —increasing effective physical density by ∼80%—and eliminates persistent indel-induced errors in the Yin–Yang codec. Furthermore, it enables a targeted resequencing strategy that achieves full recovery with 99.9% less overhead than brute-force deepening. To formalize these gains and address the lack of systematic benchmarking in the field, Polus establishes a standardized nine-metric evaluation framework that rigorously quantifies the trade-offs between reliability, density, and cost. This work provides a reproducible, quantitative foundation for next-generation, context-aware DNA storage systems.

Accurate detection of single-nucleotide variants (SNVs) and small insertions/deletions (indels) from second-generation sequencing (NGS) data is essential for clinical applications such as cancer diagnostics, infectious disease monitoring, and rapid genetic screening. However, conventional variant calling pipelines, such as GATK, decouple analysis from sequencing, deferring detection until sequencing is fully completed. We introduce RVC, a real-time variant calling framework tailored for cycle-based NGS workflows. RVC incrementally processes partially sequenced reads and continuously updates variant evidence using a scanline-based alignment algorithm and a lightweight binomial scoring model. This design enables progressive, low-latency SNVs and indels detection during sequencing, without disrupting the sequencing pipeline. In benchmark experiments using the HG002 dataset, RVC completed variant calling within tens of minutes after sequencing, significantly outperforming GATK in runtime. By tightly integrating analysis with sequencing output, RVC bridges the gap between sequencing speed and clinical responsiveness, offering a scalable and practical solution for real-time genomic diagnostics.

BACKGROUND:The development of long-read sequencing is promising for the high-quality and comprehensive de novo assembly for various species around the world. However, it is still challenging for assemblers to handle thousands of genomes, tens of gigabase-level assembly sizes, and terabase-level datasets efficiently, which is a bottleneck to large-scale de novo sequencing studies. A major cause is the read overlapping graph construction that state-of-the-art tools usually have to cost terabyte-level RAM space and tens of days for large genomes. Such lower performance and scalability are not suited to handle the numerous samples being sequenced. FINDINGS:Herein, we propose xRead, a novel iterative overlapping graph construction approach that achieves high performance, scalability, and yield simultaneously. Under the guidance of its coverage-based model, xRead converts read-overlapping to heuristic read-mapping and incremental graph construction tasks with highly controllable RAM space and faster speed. It enables the processing of very large datasets (such as the 1.28 Tb Ambystoma mexicanum dataset) with less than 64 GB RAM and obviously lower time costs. Moreover, benchmarks suggest that it can produce highly accurate and well-connected overlapping graphs, which are also supportive of various kinds of downstream assembly strategies. CONCLUSIONS:xRead is able to break through the major bottleneck to graph construction and lays a new foundation for de novo assembly. This tool is suited to handle a large number of datasets from large genomes and may play important roles in many de novo sequencing studies.

Puccinia triticina (Pt) is a heteroecious fungus needing two different plants as primary and alternate hosts throughout its life cycle. Thalictrum spp. were first identified as alternate hosts of Pt in 1921, and over 100 species have been identified. However, within China, only T. petaloideum L., T. minus L., T. minus var. hypoleucum and T. baicalense have been reported as alternate hosts of Pt. During the six-year (2015-2018, 2023-2024) field surveys in Zhangbei County (41.26°N, 115.14°E), Zhangjiakou City, Hebei Province, our research team found rust disease on T. squarrosum. This persistent infection phenomenon aroused our interest in investigating the role of T. squarrosum in the sexual reproduction of pt. To clarify whether T. squarrosum can serve as an alternate host for Pt and to analyze the source of the pathogen, this study used artificial inoculation experiments and molecular identification techniques. The results of the artificial inoculation experiments showed that the basidiospores of Pt could infect T. squarrosum, and produce pycnia on the adaxial surface of the leaf. Subsequently, aecia were produced on the abaxial side of the leaf after artificial fertilization, and the mature aecia produced aeciospores. The aeciospores were then inoculated into susceptible wheat varieties and the wheat showed typical symptoms of wheat leaf rust. These results confirmed that T. squarrosum could serve as an alternate host for Pt. For molecular identification, 20 single-aecium samples of T. squarrosum were selected. Based on sequence alignment of their ITS regions and phylogenetic analysis, it was shown that rust on T. squarrosum could be caused by infection of Pt from wheat or the species complex of P. recondita. Our study provides new insights into the sexual cycle of Pt in China and provides a scientific basis for studying the evolution of Pt virulence and optimizing control methods for wheat leaf rust.

Two-view epipolar initialization for feature-based monocular SLAM with the RANSAC approach is challenging in dynamic environments. This paper presents a universal and practical method for improving the automatic estimation of initial poses and landmarks across multiple frames in real time. Image features corresponding to the same spatial points are matched and tracked across consecutive frames, and those that belong to stationary points are identified using ST-RANSAC, an algorithm designed to detect inliers based on both spatial and temporal consistency. Two-view epipolar computations are then performed in parallel among frames and corresponding features to select the most reliable initialization. The proposed method is integrated with ORB-SLAM3 and evaluated on dynamic datasets for comparative analysis with the baseline. The experimental results demonstrate that the proposed method improves the accuracy of initial pose estimations with the construction of static landmarks while significantly reducing feature extraction scale and computational cost.

Non-noble metal electrocatalysts for the oxygen reduction reaction (ORR) are urgently needed in metal-air batteries, seawater batteries and fuel cells. Fe-N-C materials are among the most active catalysts for the ORR. Fe-N-C synthesis usually requires post-heat treatment after pyrolysis which is time-consuming and inevitably triggers inactive aggregate Fe species due to difficulties in controllable atom-level modulation. Here, highly active Fe-N-C catalysts were prepared by a simple process involving an ammonia etching treatment by using ZIF-8 as a hard template and a mixture of FeSO4 and 2-methylimidazole as the Fe, N and C precursors. The direct ammonia treatment modulates N and Fe active species and removes the unstable carbon framework to form pyrolyzed Fe-N-C nanocages with a well-dispersed pore structure. The obtained Fe-N-C exhibits a potential of 0.89 V vs. RHE at a kinetic current density of -1 mA cm-2 (E-1) for the ORR, similar to commercial Pt/C, but outperforming it in terms of stability and methanol tolerance. In situ electrochemical Raman and density functional theory provide insights into the origin of the activity of Fe-N-C materials and the underlying ORR electrocatalytic mechanisms at the molecular level.

Wheat yellow rust, caused by Puccinia striiformis f. sp. tritici (PST), is one of the most important wind-borne diseases in all wheat-growing regions, and its occurrence can lead to devastating yield losses in wheat. In China, the wheat fields in Gansu act as an inoculum reservoir during summer periods and provide PST spores for wheat in the fall. The wheat fields in Qinghai provide large amounts of oversummering inocula. The exchange and migration of spores between the two regions are essential to ensure the persistence of PST. To confirm this relationship, we studied the genetic diversity, seasonal population dynamics, role of recombination, and gene flow between PST populations in different oversummering areas of Gansu and Qinghai using molecular markers combined with a spatiotemporal sampling strategy. Shared genotypes provide molecular evidence of migration between the pathogen populations in the two regions. The distribution of genotypic frequencies indicates that the pathogen mainly flows from Qinghai to Gansu in the autumn, whereas the opposite direction of movement occurs in the spring. The inoculum source from spring wheat can be directly transmitted to autumn seedlings, not necessarily through volunteer wheat. Therefore, the bridging effect of spring wheat may play a more important role than the off-season pathogen surviving on volunteer wheat plants. Furthermore, linkage disequilibrium tests indicate that sexual recombination continues throughout the year in the Tianshui and Dingxi regions of Gansu.

The characterization of structural variants (SVs) is fundamental to genomic studies and advanced computational approaches are on demand to exert the ability of the ubiquitous high-throughput sequencing data. Herein, we propose gcSV, a read-length agnostic alignment-based approach to well-handle the issues of genome repeats, SV breakpoints and read alignments/assemblies for comprehensive, cost-effective and versatile SV calling. For long reads, its yield is 20-38% higher than state-of-the-art tools in HG002 benchmark. For hybrid sequencing, it provides a cost-effective solution (2-4x long plus 30-60x short reads) to achieve even higher yield than that of state-of-the-art tools using 30x long reads. For short reads, gcSV also achieves over 8% higher precision without any loss of sensitivity. Furthermore, gcSV confidently brings over 93,000 novel SVs comparing to the official callset of 1000 Genomes Project Phase4. The results suggest that gcSV is promising to make valuable SV discoveries in many cutting-edge studies. ### Competing Interest Statement The authors have declared no competing interest.

The advancement of high-efficiency Pt catalysts with reduced Pt loading is crucial for proton exchange membrane fuel cells (PEMFCs). This research presents a methodology that significantly increases the performance of Pt/C through the interactions between Pt and Fe-Nx/Ce-Nx on carbon, thereby effectively reducing Pt consumption. Density functional theory (DFT) calculations indicate that the presence of Fe-Nx/Ce-Nx together enhances the strong interaction between Pt and FeCe-NC, decreasing the d-band energy level (epsilon d) of Pt, which leads to the reduction of O* adsorption and acceleration of desorption at the Pt sites. Consequently, the Pt/FeCe-NC demonstrates exceptional performance for the ORR. The Pt/FeCe-NC has an E1/2 of 0.927 V and decays by only 7 mV after 30 000 accelerated stress test (AST) cycles under acidic conditions. Furthermore, the Pt/FeCe-NC (2.14 W cm-2) surpasses Pt/C (1.78 W cm-2) regarding peak power density in PEMFCs. This innovative approach clarifies the interactions between Pt and Fe-Nx/Ce-Nx, providing a framework for the design of advanced catalysts.

Long-read sequencing technologies have great potential for the comprehensive discovery of structural variations (SVs). However, accurate genotype assignment for SVs remains challenging due to unavoidable sequencing errors, limited coverage, and the complexity of SVs. Herein, we propose cuteFC, which employs self-adaptive clustering along with a multiallele-aware clustering to achieve accurate SV regenotyping through a force-calling approach. cuteFC also applies a Genome Position Scanner algorithm to improve its application efficiency. Benchmarking evaluations demonstrate that cuteFC outperforms state-of-the-art methods with 2-5% higher F1 scores and constructs a higher-quality genomic atlas with minimal computational resources. cuteFC is available at https://github.com/Meltpinkg/cuteFC and https://zenodo.org/records/14671406 .

Visual-based Simultaneous Localization and Mapping has been extensively studied and applied to navigation in unmanned aerial vehicles. However, SLAM remains challenging in dynamic environments, where moving objects may be misidentified as static landmarks, compromising pose estimation. While many data-driven solutions have been proposed to address this issue, they often come at the cost of real-time performance, particularly on resource-constrained platforms. This paper presents a geometry-based approach to prune non-stationery features with minimal computational overhead. Features corresponding to the same spatial point are matched and tracked across consecutive frames, with epipolar errors computed over multiple frame pairs. Features exhibiting epipolar consistency over time are classified as stationary features and are constructed into map points. The proposed method is implemented into the state-of-the-art ORB-SLAM3 framework for evaluation. Dataset experiments demonstrate that the proposed SLAM system outperforms the baseline in both mapping efficacy and localization accuracy, while maintaining real-time performance without GPU acceleration. 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/)

Atomically dispersed metal-nitrogen-carbon (M-N-C) catalysts have exhibited encouraging oxygen reduction reaction (ORR) activity. Nevertheless, the insufficient long-term stability remains a widespread concern owing to the inevitable 2-electron byproducts, H2O2. Here, we construct Co-N-Cr cross-interfacial electron bridges (CIEBs) via the interfacial electronic coupling between Cr2O3 and Co-N-C, breaking the activity-stability trade-off. The partially occupied Cr 3d-orbitals of Co-N-Cr CIEBs induce the electron rearrangement of CoN4 sites, lowering the Co-OOH* antibonding orbital occupancy and accelerating the adsorption of intermediates. Consequently, the Co-N-Cr CIEBs suppress the two-electron ORR process and approach the apex of Sabatier volcano plot for four-electron pathway simultaneously. As a proof-of-concept, the Co-N-Cr CIEBs is synthesized by the molten salt template method, exhibiting dominant 4-electron selectively and extremely low H2O2 yield confirmed by Damjanovic kinetic analysis. The Co-N-Cr CIEBs demonstrates impressive bifunctional oxygen catalytic activity (▵E=0.70 V) and breakthrough durability including 100 % current retention after 10 h continuous operation and cycling performance over 1500 h for Zn-air battery. The hybrid interfacial configuration and the understanding of the electronic coupling mechanism reported here could shed new light on the design of superdurable M-N-C catalysts.