Pt based materials are the most efficient catalysts for hydrogen evolution reaction (HER). However, the high cost and low earth abundance severely retard its commercial applications. Moreover, degradation of noble metal materials inevitably occurs during long-term HER process, especially in acidic media. Herein, core-shell structured high-index facet PtNi nanocrystals (HIF-PtNi NCs) coated with a polydopamine derived carbon nanofilm (cPDA) were fabricated for HER with enhanced catalytic activity and significantly improved stability to minimize Pt loadings. HIF-PtNi NCs possess a uniform concave morphology with average diameter of ∼42 nm in which the predominantly exposed crystal lattices spacing is 0.196 nm that is attributed to the (200) planes. The exposed high-energy surfaces exhibit better HER performance than low-energy surfaces such as commercial Pt/C catalyst and can reduce Pt loadings for commercial applications. Experiments and density functional theory (DFT) calculations confirmed that the amorphous carbon nanofilm significantly inhibited the deformation and degradation of the HIF-PtNi NCs catalyst in the core. HIF-PtNi NCs coated with cPDA nanofilm (HIF-PtNi NCs/cPDA) exhibits the HER overpotentials of 42 mV and the Tafel slope of 36.3 mV dec-1 at current density of 10 mA cm-2 with a low Pt loading of 140 μg cm-1. HIF-PtNi NCs/cPDA shows almost no decay in HER activity after an accelerated durability test that conducted at a current density of 150 mA cm-2 for 100 h in 0.5 M H2SO4 solution.

Conventional heading-keeping autopilot logic exhibits well-known performance limitations under complex route geometry and environmental disturbances. Motivated by this limitation, this paper proposes a line-adhering intelligent navigation framework for disturbance-aware path-following of autonomous ships. The core idea is based on numerical simulation scenarios representing curved inland/coastal routes under wind- and current-disturbance conditions. The addressed gap lies in the limited integration of route-geometry adherence, human-like maneuvering logic, and disturbance-aware controller reconfiguration within conventional heading-centered ship path-following frameworks. Therefore, a rough-set classifier identifies disturbance modes and reconfigures PID, LQR, and MPC controllers in real time. Moreover, a vessel-dynamics constrained Bézier refinement method generates high-resolution reference paths aligned with navigational curvature limits. Mathematical models including the Nomoto and MMG formulations are incorporated to ensure controllability and dynamic feasibility. Results show that the proposed framework improves path-following precision, robustness, and comfort under the considered simulation conditions.

Data classifiers aim to assign the sample data with unknown class into a category with a label of specific class. Due to the complexity of sample data distributions and the correlations among attributes, it is inevitable that sample data of different classes overlap partially or completely in attribute space. These indistinguishable samples (IDSs) in overlapping regions are an important cause of misclassification. To solve this, this paper proposes a novel coarse-fine classifier via evidential reasoning (ER) rule under the framework of evidence theory. Firstly, in the pre-classification stage, a coarse classifier is designed to perform preliminary classification for capturing IDSs within the attribute space and then a relabeling strategy is presented to measure the uncertainty of IDSs by assigning non-unique labels with a certain belief degree to IDS. Secondly, in the formal classification stage, a fine classifier is designed to uniformly and finely model IDSs and other normal sample data as multiple pieces of evidence. The ER rule is employed to combine these activated evidence. Its inherent orthogonal sum operation focuses the combined belief degree onto the true class label, thereby reducing misclassification. Finally, Experiments on benchmark datasets and a gearbox fault diagnosis application validate the method's effectiveness and superiority.

In this paper a novel monocular 3D object detection algorithm, namely MonoStrong, is constructed as a new solution to enhance the navigation safety for general and autonomous ships in particular with the assistance of visual detections. Initially, a channel attention-enhanced feature network is designed to adaptively strengthen feature channels corresponding to critical geometric structures, improving the representation of complex ship characteristics, particularly for feature-sparse ships at medium and long distances. Subsequently, a multi-scale feature fusion module with attention mechanisms is formulated to enhance the robustness of perception through cross-level feature interaction, further strengthening feature representation for ships with scale variations. An optimized 2D-3D bounding box association loss function is then constructed to effectively reduce errors caused by rapid heading variations at short distances and localization drift at extended ranges. A multi-scenario 3D visual dataset for ships, called Kitti3D-Ship, is constructed to validate the proposed technique. The results demonstrate that MonoStrong outperforms existing methods in ship attitude estimation, particularly in longrange, high-density, and multi-scale scenarios. The method offers a reliable solution for intelligent maritime perception and contributes to the safe operation of ships.

This article investigates the problem of robust H infinity auxiliary observer (RHAO)-based nonlinear fuzzy fault-tolerant steering control for unmanned marine surface vessels (UMSVs) with sensor fault detection and on-demand compensation. First, an RHAO is developed for the simultaneous estimation of unknown states and sensor fault. By introducing a novel auxiliary state, the unknown term in the RHAO is eliminated, and a practically realizable RHAO is derived, which significantly enhances the observation accuracy. Subsequently, utilizing the obtained observation information, an RHAO-based nonlinear fuzzy fault-tolerant controller (NFFTC) with on-demand compensation is proposed, which stabilizes the nonlinear steering control system while reducing conservatism. Meanwhile, real-time analysis of residual signal is performed, and an adaptive threshold and energy-driven fault detection scheme is designed to ensure timely detection and accurate warning when a sensor fault occurs, which provides an early warning symbol for invoking the on-demand compensation strategy. The effectiveness and feasibility of the developed steering control strategy are finally verified via simulation studies.

Intelligent navigation decision support systems are crucial for maritime safety, yet these systems frequently exhibit limited adaptability and reliability in novel, non-predefined scenarios, constituting a persistent challenge. This study proposes Navigation-GPT, a dual-core large language model (LLM) agent designed for intelligent marine navigation. The framework leverages the strong generalization capability of LLMs in unfamiliar situations. It employs a large-scale LLM with ReAct prompting as its control core, responsible for task parsing, planning, and orchestrating external tools to mitigate hallucinations. Furthermore, we fine-tune a lightweight LLM in two stages: using LoRA and a novel rule-controlled GRPO (RC-GRPO) method to develop a specialized agent decision core. This core generates COLREGs-compliant high-level collision avoidance decisions, which are translated into dynamically feasible reference trajectories using a ship dynamics model formulated according to Fossen's equations. A PID-based controller then tracks these trajectories to guide the ship through the resulting avoidance maneuver. Experimental results show that Navigation-GPT completes the process from task reception to decision output in 11.13 s, remaining within the critical safety window for collision avoidance, though longer than the 0.73 s of traditional methods. In complex scenarios, it achieves an 86% collision avoidance success rate and a 90% behavioral compliance rate, outperforming its base model Qwen2.5-7B by 38% and surpassing benchmarks including the dynamic window approach, artificial potential field, and other LLMs (Qwen2.5-0.5B, Qwen2.5-14B, DeepSeek, GPT-4o). This work integrates LLM technology with traditional navigation systems, offering a comprehensive solution that enhances both safety and operational efficiency across diverse maritime scenarios.

Domain generalization is a critical requirement in industrial fault diagnosis, as models must generalize reliably to unseen and unknown target domains. However, most existing approaches implicitly assume balanced fault distributions across domains, neglecting the severe class imbalance prevalent in real industrial data. In practice, fault samples are long-tailed and unevenly distributed across multiple domains, giving rise to the more challenging problem of long-tailed multi-domain generalization, where domain shift and class imbalance are inherently coupled. In this paper, we move beyond reweighting and propose a discriminative contrastive learning framework for long-tailed multi-domain generalization in intelligent fault diagnosis. We formulate a new long-tailed multi-domain generalization task and design a novel loss function that integrates supervised classification with contrastive learning to enhance class separability under severe imbalance. We further introduce cross-domain intraclass consistency regularization to explicitly enforce consistency among samples of the same fault class across different domains. By modeling discriminative relationships across domains, the proposed framework effectively mitigates long-tailed bias and improves generalization to unseen target domains. Extensive experiments on multiple benchmarks demonstrate clear and consistent advantages over state-of-the-art reweighting-based and domain generalization methods, particularly in recognizing minority fault classes under significant domain shifts.

Addressing the challenges posed by low Signal-toNoise Ratio, weak texture, and strong clutter interference in maritime radar imagery, this paper proposes a performance evaluation framework for object detection and Multi-Object Tracking based on YOLOv8. The study aims to systematically verify the applicability of the "Tracking-by-Detection" paradigm in maritime radar scenarios and to deeply investigate the practical value of appearance features in data association. Leveraging a self-constructed dataset of 1,816 high-precision annotated radar images, this paper first optimized and trained a YOLOv8 detector tailored to radar characteristics, establishing a high-performance detection baseline with a mAP50 of 99.42%. On this basis, a comparative evaluation was conducted on over ten mainstream tracking algorithms, including SORT, ByteTrack, and SparseTrack. In particular, addressing the lack of visual details in radar targets, this paper constructed a radar-specific Reidentification dataset and trained a multi-scale OSNet model for ablation studies.

This article addresses the fuzzy adaptive fault-tolerant path following control (PFC) problem of unmanned surface vessels (USV) with side-slip compensation, unmeasurable states, and actuator fault. The indirect control strategy is adopted to decompose the PFC into a guidance system and a heading control system. For the guidance system, an adaptive line-of-sight (ALOS) guidance with a time-varying lookahead distance, which yields the desired heading by side-slip estimation and compensation, is designed, thus transforming the PFC problem into the heading tracking control problem. Meanwhile, for the heading control system, a fuzzy state observer is first developed to estimate the unmeasurable states. Next, an adaptive scheme is designed to estimate the time-varying actuator fault and an adaptive fault-tolerant controller is designed by combining the state and fault estimation to enhance the fault-tolerance of PFC system and ensure the performance of path tracking control. Then, the observer and controller are designed integrally, and the stability is analyzed using the integrated design principle. Finally, the simulation case study and comparison experiments of PFC for USV are conducted to demonstrate the performance of this method.

In the domain of autonomous ship navigation, the construction of bird’s-eye view (BEV) layouts for waterways has obvious significance. A helmsman can generate the BEV layout of the waterway using his/her eyes only. To simulate this intelligence, a novel neural network-based algorithm named Waterway-BEV is proposed, which enables reconstructing a local map formed by the waterway layout and ship occupancies in the bird’s-eye view given a first person view monocular image only. Waterway-BEV employs an efficient SEResNeXt encoder to extract features from first person view (FPV) monocular images, capturing deep semantic information related to waterways and ships. Due to the variations in information across different perspectives, Waterway-BEV incorporates a Cross-View Transformation Module, which takes the constraint of cycle consistency between views into account and makes full use of their correlation to strengthen the view transformation and scene understanding. To fully leverage the feature output of the SEResNeXt encoder, Waterway-BEV employs a decoder based on a dedicated lightweight network. This decoder is responsible for decoding the enhanced bird’s-eye view (BEV) feature maps and generating the BEV layout. By employing the Focal Loss as the loss function for model optimization, Waterway-BEV takes into account the quantity and classification difficulty of ship samples during the training process, thereby improving the generation performance and convergence speed. The experiments demonstrated that Waterway-BEV achieved notable performance metrics, with mIOU and mAP rates reaching 97.8% and 98.2%, respectively, in waterway bird’s-eye view layout generation. Waterway-BEV outperformed other state-of-the-art (SOTA) algorithms in generating BEV layouts of waterways. In particular, during specialized scenarios such as crossroads of waterways and tasks involving small target ships, Waterway-BEV consistently generated satisfactory bird’s-eye view layouts, demonstrating robustness and applicability.

Existing navigation decision support systems often perform poorly when handling non-predefined navigation scenarios. Leveraging the generalization capabilities of large language model (LLM) in handling unknown scenarios, this research proposes a dual-core framework for LLM applications to address this issue. Firstly, through ReAct-based prompt engineering, a larger LLM core decomposes intricate navigation tasks into manageable sub-tasks, which autonomously invoke corresponding external tools to gather relevant information, using this feedback to mitigate the risk of LLM hallucinations. Subsequently, a fine-tuned and compact LLM core, acting like a first-mate is designed to process such information and unstructured external data, then to generates context-aware recommendations, ultimately delivering lookout insights and navigation hints that adhere to the International Regulations for Preventing Collisions at Sea (COLREGs) and other rules. Extensive experiments demonstrate the proposed framework not only excels in traditional ship collision avoidance tasks but also adapts effectively to unstructured, non-predefined, and unpredictable scenarios. A comparative analysis with DeepSeek-R1, GPT-4o and other SOTA models highlights the efficacy and rationality of the proposed framework. This research bridges the gap between conventional navigation systems and LLMs, offering a framework to enhance safety and operational efficiency across diverse navigation applications.

Traditional discriminative correlation filters used for tracking unmanned aerial vehicle objects are often disrupted by concealed noise, resulting in unstable tracking results. Various methods have been developed to search for optimal feature combinations and construct feature weight pools. However, these methods often overlook the significance of different feature channels in tracking frames. Irrespective of the availability of the effective target information, a tracker regards all feature channels similarly. This makes it challenging for the tracker to avoid learning the background noise from such feature combinations. This study proposes a channel-level feature-weighting method called learning feature-weighted regularization discriminative correlation filters (FWRDCF). By introducing feature-weighted regularization (FWR) that automatically adjusts the weights of the feature channels into each frame, the FWRDCF tracker can significantly suppress background noise. Furthermore, the alternating direction method of multipliers is used to obtain the closed-form solution of the model, thereby establishing a robust correlation filter-tracking architecture. Experiments on UAV123@10fps, UAV123, DTB70, and UAVDT demonstrated that the FWRDCF tracker achieved better tracking performance than 15 other state-of-the-art trackers. An integration study of three baselines (AutoTrack, STRCF, and BACF) reveals that the proposed FWR can be integrated with trackers with multi-channel features.

Syngas (H2/CO) is an essential chemical feedstock for industrial products. In these focal points, electrocatalytic CO2 reduction has emerged as a desirable strategy for realizing effective syngas production to satisfy energy and environmental requirements. In this work, a metal-molecule hybrid electrode with inherent H2 generation favorability has been crafted by loading molecular Co(Ni)-bpy (bpy = 2,2'-bipyridine) complexes on Ag foil. The efficient and stable CO2-to-CO conversion with adjustable faradic efficiency from 13 to 98% was realized by optimizing the Co(Ni)-bpy complexes. The regulation of molecular catalysts with the merits of high electron affinity can provide a coordination environment that allows for the localization of Co/Ni active sites at optimal positions with lower binding energies, maintaining their monodisperse properties, and being beneficial for strengthening the CO2 binding and inhibiting competitive reactions. An in-depth understanding of surface and coordination status has been realized by FIB-HRTEM and EXAFS, which confirm that the intimate metal-molecular interaction and well-dispersed mononuclear Co/Ni active sites play vital roles in enhancing catalytic performance. The strong electron residual between the Ag surface and metal-coordinated molecular catalysts may also contribute to the dramatic CO2-to-CO conversion. This study highlights the beneficial role of metal-molecule interactions in electrocatalytic reactions and contributes to ongoing efforts toward achieving controllable selectivity in electrocatalytic reduction of CO2 to syngas using molecular catalysts.

Silicon (Si) anodes, with a theoretical specific capacity of 4200 mAh g-1, hold significant promise for the development of high-energy-density lithium-ion batteries (LIBs). However, practical applications are hindered by sluggish charge transfer kinetics, substantial volume expansion, and an unstable solid electrolyte interphase during cycling. To address these challenges, we propose a centimeter-scale Si anode design featuring a three-dimensional continuous network structure of Si nanowires (SiNWs) decorated with high-density Ag nanoparticles (Ag-SiNWs-Net) on both the surface and internally. This architecture effectively mitigates mechanical stress from Si volume changes through the high-aspect-ratio wire network. Additionally, the distribution of Ag nanoparticles on the Si induces electronic structure redistribution, generating built-in electric fields that accelerate charge transfer within the Si, significantly enhancing rate performance and cycling stability. The Ag-SiNWs-Net anode achieves a high reversible capacity of 3780.9 mAh g-1 at 0.1 A g-1, with an initial coulombic efficiency of 85.1 %. Moreover, the energy density of full cells assembled with Ag-SiNWs-Net anodes and LiFePO4 cathodes can be pushed further up to 395.8 Wh kg-1. This study offers valuable insights and methodologies for the development of high-capacity and practical Si anodes. (c) 2025 Published by Elsevier Ltd on behalf of The editorial office of Journal of Materials Science & Technology.

A control method based on proximal policy optimisation (PPO) algorithm is proposed to control autonomous berthing of unmanned vessels. Firstly, Bessel curves are used to generate the desired path, and the heading error, speed error, and lateral error are obtained by calculating with the current state of the ship, and then these errors are used as inputs to establish the reward function mechanism, and then the state of the ship and the reward obtained are inputted into the PPO algorithm so that the intelligent body can take behaviours according to the state and the reward, and finally, the actions of the intelligent body are inputted into the ship model to achieve bi-variable control of the ship’s rudder angle and thrust. Finally, the actions of the intelligent body are inputted into the ship model to achieve the bivariate control of the rudder angle and thrust. The experimental results show the effectiveness of PPO reinforcement learning in the automatic berthing process of unmanned vessels with low speed, interference and low rudder efficiency.

This paper proposes a hierarchical reasoning method for shoreline extraction that relies solely on a camera and visual algorithms. First, the DeepLabV3+ algorithm is used to extract water bodies from images, and the Canny operator is applied to outline these water bodies. Subsequently, the YOLOv8-seg algorithm is used to extract contours of objects encountered in the image, helping to eliminate occlusions caused by these objects. Finally, the KeyPoint-RCNN algorithm is employed to distinguish between the left and right shorelines, achieving the extraction of the shoreline. For key point detection, a creation method of datasets is proposed that improves the target and key point detection capabilities of the KeyPoint-RCNN algorithm by 27.1

Aberrant or non-standard operations by ship drivers are a leading cause of water traffic accidents, making the development of real-time and reliable behavior detection systems critically important. However, the environment within a ship’s bridge is significantly more complex than typical scenarios, such as vehicle driving or general security monitoring, which results in poor performance when applying generic algorithms. In such settings, both the accuracy and efficiency of existing methods are notably limited. To address these challenges, this paper proposes a cross-modal behavioral intelligence framework designed specifically for a ship’s bridge, integrating multi-target tracking, behavior recognition, and feature object association. The framework employs ByteTrack, a high-performance multi-object tracker that maintains stable tracking even when subject to occlusions or motion blur through its novel association mechanism, using both high and low confidence detection boxes, for multi-driver tracking. Combined with an improved Temporal Shift Module (TSM) algorithm for behavior recognition, which effectively resolves issues concerning target association and action ambiguity in complex environments, the proposed framework achieves a Top-1 accuracy of 82.1%, based on the SCA dataset. Furthermore, the method incorporates a multi-modal decision optimization strategy, based on spatiotemporal correlation rules, leveraging YOLOv7-e6 for simultaneous personnel and small object detection, and introduces the Accuracy of Focused Anomaly Recognition (AFAR) metric to enhance the anomaly detection performance. This approach improves the anomaly detection rate, up to 81.37%, with an overall accuracy of 80.66%, significantly outperforming single-modality solutions.

To address the prevalent challenges of small target scales and occlusions encountered in port and ship operations scenarios, this work proposes the YOPR algorithm, which builds upon the YOLO framework. To tackle the aforementioned practical constraints, the following enhancements were introduced. Firstly, the feature extraction network was augmented with multiple ParNet Attention mechanisms. This improvement serves to optimize the deep convolutional layers’ capability to extract features from small-scale targets. Secondly, during the feature fusion process, the high-level convolutional outputs were replaced with those from shallower convolutions, enabling the retention of higher-resolution feature maps and thus more effective features for small targets. Evaluated on the self-constructed Person3100 dataset, the YOPR algorithm demonstrated satisfactory adaptability and robustness compared to other models, while meeting the real-time computational requirements of actual port monitoring systems.

The new mode of ship formation navigation can significantly enhance the level of ship intelligence and improve navigation efficiency. To address the obstacle avoidance problem of ship formations under obstacle-constrained environments, a control model for Autonomous Surface Vessel (ASV) formation is constructed using the virtual structure method, decomposing the formation control problem into path tracking control and path parameter coordination. Considering both dynamic/static obstacles during the navigation of ASVs formation, an obstacle avoidance control strategy based on artificial potential fields is proposed. It achieves formation maintenance and obstacle avoidance control for ASVs in complex environments. Simulation results show that ASVs can track paths in desired formation and effectively avoid obstacles.