Parking-line segmentation for automated parking in underground garages is challenged by sparse targets, perspective-compressed thin structures, and visually similar disturbances such as cracks, joints, uneven illumination, and reflections. We propose a lightweight YOLO26n-sem framework with Semantic-Preserving Shallow Residual Detail Injection (SRDI) and Interface-Preserving Deep Semantic Slimming (IPDS). SRDI injects scale-aligned P2 details into the P3 pathway through a bounded learnable residual coefficient, while IPDS reduces hidden widths in deep P5 modules without changing the external P5 interface or top-down propagation topology. On a four-class bird’s-eye-view dataset acquired by a vehicle-mounted surround-view system, the model achieves 85.966% mIoU, 83.081% Parking IoU, and 87.700% Thin Recall, improving YOLO26n-sem by 0.500, 0.973, and 1.556 percentage points, respectively. Parameters decrease by 24.9%, from 1.631 M to 1.225 M, and latency decreases from 8.8203 to 7.0321 ms (142.21 FPS). Zero-shot evaluation on an independent public underground-garage dataset achieves 77.837% mIoU and 72.625% Parking IoU, exceeding the baseline by 1.416 and 2.917 percentage points, respectively, without target-domain adaptation. Overall, the proposed model improves segmentation accuracy while reducing parameter count and inference latency, and its advantage over the baseline remains observable under the tested cross-garage domain shift.

Hydrogen peroxide (H₂O₂) is widely used in semiconductor cleaning and etching, where ultralow levels of metallic, anionic, organic, and particulate impurities must be strictly controlled. Industrially produced H₂O₂ therefore requires extensive downstream purification before it can meet electronic-grade specifications. Conventional purification routes based on distillation or rectification, adsorption, ion exchange, and final filtration are technically mature, but they remain constrained by substantial energy consumption, multiple treatment stages, chemical regeneration, secondary waste generation, and safety risks associated with H₂O₂ decomposition. This review critically evaluates membrane technologies for purifying and concentrating electronic-grade H₂O₂. Microfiltration and ultrafiltration are discussed as front-end clarification processes, nanofiltration as an intermediate impurity-load-reduction step, and reverse osmosis as the membrane process with the strongest direct experimental for ionic-impurity removal from concentrated H₂O₂. Pervaporation and membrane distillation are assessed as emerging water-removal technologies, although their industrial applicability remains insufficiently validated. Membrane material strategies, including oxidation-resistant polymers, inorganic and hybrid membranes, antioxidant-containing composites, and emerging MOF- and two-dimensional-material-based membranes, are also evaluated. Particular attention is paid to the limited direct evidence available for emerging materials and to the risks of H₂O₂ decomposition, material leaching, particle release, and deterioration of membrane selectivity. The available evidence indicates that membrane processes are currently more appropriately regarded as complementary clarification, purification, polishing, or concentration units rather than complete replacements for established industrial technologies. Future studies should prioritize long-term oxidative stability, ppb- and ppt-level impurity validation, low H₂O₂ loss, module-material compatibility, process safety, and continuous pilot-scale techno-economic assessment.

To address the issues of low adaptability and significant tracking errors in parking scenarios when using fixed look-ahead distance Pure Pursuit (PP) algorithms, this paper proposes an automatic parking path tracking control algorithm based on Fuzzy Pure Pursuit (FPP). Considering the influence of road curvature on look-ahead distance, a fuzzy controller is designed to output speed proportionality coefficient and curvature proportionality coefficient. This enables adaptive adjustment of the look-ahead distance according to vehicle speed and road curvature, thereby enhancing path adaptability and tracking accuracy. Prescan/CarSim/Simulink simulation results demonstrate that in vertical parking scenarios, the FPP-based tracking control algorithm outperforms traditional PP algorithms in tracking performance for desired paths and heading angles. The tracking error is reduced by 4.8%, and the heading angle error is reduced by 7.3%. The test results of the Apollo advanced platform show that, under different initial heading angles, the vehicle is able to successfully track the parking path and completes the parking operation without collisions. The tracking control algorithm based on FPP has excellent environmental adaptability.