The precise identification of safe navigable areas constitutes a foundational requirement for autonomous driving in complex unstructured environments. Existing studies have predominantly concentrated on extracting local or single-scale features, neglecting the substantial complementarity between multi-scale representations in complex natural environments, which constrains semantic understanding. To address this issue, a Semantic-Aware Terrain Segmentation Network (SATSNet) is proposed for navigable area recognition with guidance-entropy, which incorporates global sparse contextual information as prior guidance and progressively explores the latent complementarity among multi-scale features from both semantic hierarchy and spatial structure perspectives, achieving discriminative terrain feature aggregation. SATSNet is built around two essential modules: the Entropy-Guided Global Sparsification Module (EGGSM) and the Semantic-Spatial Fusion Module (SSFM). EGGSM performs selective channel-wise global semantic modeling through entropy-guided sparsification, providing representative long-range dependency semantic cues for guidance. SSFM comprises the Attention-Entropy Fusion Unit and the Dynamic Weight Allocation Unit, which leverage entropy-based attention to explore the semantic-spatial complementarity among adjacent features, facilitating the aggregation of semantic-aware terrain features. Validation on multiple challenging wild datasets, as well as large-scale street-view and real-world application datasets, demonstrates that SATSNet outperforms existing state-of-the-art methods in navigable area segmentation, providing a robust solution for safe autonomous navigation.
To optimize the photocatalytic efficacy of bismuth oxybromide (BiOBr), a series of BiOBr/CuInS2 (BCIS-x) heterojunctions were synthesized via a facile solvothermal approach, wherein CuInS2 nanoparticles were uniformly dispersed on the surface of BiOBr nanosheets, forming a sesame-cookie-like structure. The systematic characterization results demonstrated that the formation of the BiOBr/CuInS2 heterojunction effectively modulated the band structure, resulting in a narrowed band gap, extended visible light absorption range, and the significant improvement of the charge carrier separation efficiency. These synergistic effects collectively contributed to the enhanced photocatalytic performance of the CO2 reduction reaction. Notably, the optimized BCIS-3 exhibited superior photocatalytic activity, achieving the highest CO yield of 9.63 mu mol g-1 h-1 under simulated sunlight irradiation for 2 h, which represents 4-fold enhancement compared to pristine BiOBr. This remarkable improvement for photocatalytic CO2 reduction performance can be attributed to the effective charge transfer at the heterojunction and the inhibition of photogenerated electron-hole pair recombination at the heterojunction interface.
The fifth edition handbook of FMEA introduced action priority (AP) to address issues caused by difference in the weightings of risk factors severity (S), occurrence (O), and detection (D). However, it omitted the underlying design theory of AP, limiting interpretability and scenario-specific adaptation. This study proposes a flexible AP design methodology incorporating customizable risk factor weights and adjustable failure mode risk levels. First, a risk matrix for O and D is constructed for different S risk levels, and a risk index based on factor-weighted distance is defined. A risk utility function then models the risk preference differences, generating a stratified distribution of S risk levels, which determines failure mode risk levels and corresponding AP assignments. Applying this method with weights of 0.5, 0.3, 0.2 for S, O, D, and a Type-A index distribution yields results fully consistent with the handbook. Furthermore, an enhanced AP is proposed, subdividing the three-category AP (High, Medium, and Low) into 125 distinct levels. Finally, the method is developed to process failure mode assessment in aviation manufacturing. A case study shows that a refined five-category AP (Extremely High to Extremely Low) improves discrimination among failure modes within the same category and enhances practical applicability.
The prevailing rheological master curve models and time-temperature superposition techniques were not originally developed with consideration of physical hardening. Consequently, their applicability and sensitivity under such conditions remain unclear and warrant further investigation. This study presents the first systematic evaluation of the applicability and sensitivity of three master curve models (CAM, GLSM, and 2S2P1D) and five kinds of the time-temperature superposition techniques (Arrhenius, WLF, Kaelble, Modified Kaelble, and Non-Linear Least Squares) under physical hardening effects. Quantitative assessments were conducted by performing frequency sweeps on base asphalt, SBS-modified asphalt, and asphalt mastic under two low-temperature storage modes, followed by the construction of complex modulus master curves. The goodness-of-fit for the time-temperature superposition techniques followed the order: Arrhenius < WLF < Kaelble < Modified Kaelble < Nonlinear least squares, with the latter three demonstrating superior fitting performance. In contrast, the combination of the CAM model and Arrhenius yielded a R2 of only 0.911, revealing significant limitations. Based on the principles of optimal goodness-of-fit and conservative safety, the combination of the GLSM and the Modified Kaelble is recommended for constructing rheological master curves of asphalt materials under physical hardening. The findings of this study provide critical references and theoretical support for the accurate selection and effective characterization of asphalt material performance in cold regions.
The elevational distribution of microbial communities in the surface soils of natural lake wetlands has been widely studied. However, it remains unknown whether microbiota-mediated carbon metabolism in different soil layers exhibits a similar change pattern with elevation. In this study, we analyzed changes in functional genes that regulate carbon cycling in different soil layers and elevations using metagenomic analysis in Poyang Lake, China. The genes associated with the carbon cycle, particularly those involved in extracellular and intracellular carbon decomposition, are more abundant in surface soil (0–20 cm) than in deeper soils (40–60 cm and 80–100 cm). The impact of elevation on the soil carbon cycle in wetlands varies across different soil layers. In surface soil, only a gradual increase in the abundance of methane oxidation genes with increasing elevation was observed. In deeper soils, the abundance of fermentation (formate, lactate metabolism) and anaerobic carbon fixation genes decreased with elevation, while methane oxidation (mmo, pmo) and CO oxidation (coxMLS) genes exhibited opposite trends. As soil depth increased, the genes involved in the soil carbon cycle exhibited stronger interrelationships and network complexity, and a greater number of carbon cycle genes showed a strong correlation with soil organic carbon. Furthermore, surface soil organic carbon was primarily influenced by plant characteristics. Our findings suggest that microorganisms, rather than plants, play a crucial role in deep soil carbon storage and transformation.
