Accurate prediction of Arctic sea ice concentration (SIC) is critical for understanding climate change and ensuring safe Arctic shipping. However, conventional U-Net architectures often degrade deep features and lack explicit physical constraints. To address this, we propose a physics-guided U-Net with attention residual connections, in which physical forcing steers feature learning at the architecture level rather than entering solely as loss-level constraints. An attention residual mechanism enables adaptive multi-scale feature fusion across layers, while a physics-constrained module is embedded throughout the encoder–decoder to enforce physical consistency. Thermodynamic and dynamic variables (sea surface temperature, 2 m air temperature, and 10 m wind components) are fused via cross-modal attention and physics-guided skip connections. A dual-input framework captures local fine-scale ice edge details and large-scale Arctic background fields simultaneously. The composite loss combines a weighted Charbonnier term and a freezing loss to better penalize ice edge mismatches and enforce freeze/melt physics. Experiments using National Snow and Ice Data Center (NSIDC) and ERA5 data demonstrate that our model outperforms six representative deep learning baselines at the 1-month lead under a unified multi-seed protocol, with all pairwise improvements statistically significant, and retains substantial skill at a 6-month lead, accurately reproducing the September minimum, with ice edges closely tracking observations. These results confirm that the physics-guided attention residual U-Net enhances spatial fidelity and physical consistency of subseasonal-to-seasonal SIC predictions, thereby supporting robust, medium-to-long-term assessment of Arctic navigability windows.

This paper proposes a dynamic obstacle-avoidance algorithm for unmanned surface vehicles (USVs) that combines a Fuzzy-enhanced Dynamic Window Approach (FDWA) with an Improved Bidirectional A*–Improved Grey Wolf Optimizer (IBA*–IGWO) framework. Firstly, the traditional dynamic window method (DWA) is improved by adopting an initial heading angle optimization strategy to reduce the heading deviation of unmanned vessels during cruising. Secondly, a fuzzy controller is introduced, which can adaptively adjust the weight coefficients in the cost function of the DWA algorithm based on the current position of the unmanned vessel, surrounding environmental information, etc., to improve obstacle avoidance ability and adaptability in different environments. Finally, using the global static cruise route provided by the IBA*–IGWO algorithm, key nodes are selected as local endpoints for the FDWA algorithm to ensure that the unmanned vessel can perform cruise tasks according to the optimal plan during navigation and make dynamic adjustments in case of emergencies. The simulation results demonstrate the feasibility of the proposed method in handling unknown and dynamic obstacles under the current grid-based experimental settings, while enabling the USV to return to the pre-planned global route after local obstacle avoidance. These results provide a basis for further development toward more robust and rule-aware autonomous navigation in realistic maritime environments.