The growing use of remote-sensing technologies has placed greater demands on object-detection algorithms, which still face challenges. This study proposes a hierarchical adaptive feature aggregation network (HAF-YOLO) to improve detection precision in remote-sensing images. It addresses issues such as small object size, complex backgrounds, scale variation, and dense object distributions by incorporating three core modules: dynamic-cooperative multimodal fusion architecture (DyCoMF-Arch), multiscale wavelet-enhanced aggregation network (MWA-Net), and spatial-deformable dynamic enhancement module (SDDE-Module). DyCoMF-Arch builds a hierarchical feature pyramid using multistage spatial compression and expansion, with dynamic weight allocation to extract salient features. MWA-Net applies wavelet-transform-based convolution to decompose features, preserving high-frequency detail and enhancing representation of small-scale objects. SDDE-Module integrates spatial coordinate encoding and multidirectional convolution to reduce localization interference and overcome fixed sampling limitations for geometric deformations. Experiments on the NWPU VHR-10 and DIOR datasets show that HAF-YOLO achieved mAP50 scores of 85.0% and 78.1%, improving on YOLOv8 by 4.8% and 3.1%, respectively. HAF-YOLO also maintained a low computational cost of 11.8 GFLOPs, outperforming other YOLO models. Ablation studies validated the effectiveness of each module and their combined optimization. This study presents a novel approach for remote-sensing object detection, with theoretical and practical value.

This article investigates a prescribed-time trajectory tracking control strategy for USVs considering marine environmental interferences and unmodeled dynamics. Firstly, a fixed-time extended state observer is introduced to quickly and accurately observe the compound perturbations including ocean disturbances and unmodeled dynamics. Subsequently, a prescribed-time prescribed performance function is utilized to obtain guaranteed transient performance within a predefined time. Finally, combining the fixed-time extended state observer, dynamic surface control technique, and prescribed-time prescribed performance control, a prescribed-time prescribed performance control strategy is developed to guarantee that the tracking errors converge to a predefined performance constraint boundary within a prescribed time. The effectiveness and superiority of the presented control strategy is verified by the simulation results.

Formation obstacle avoidance is an essential attribute of the cooperative task in unmanned surface vehicle (USV) formation. In real-world scenarios involving multiple USVs, both formation obstacle avoidance and formation recovery after obstacle avoidance play a critical role in ensuring the success of collaborative missions. In this study, an Interfered Fluid Dynamic System (IFDS) algorithm was used for obstacle avoidance due to its excellent robustness, high computational efficiency and path smoothness. The algorithm can provide good local path planning for USVs. However, the use of the IFDS on USVs still has the defect of local extreme values, which has been effectively modified to obtain an enhanced IFDS (EIFDS). In formation, based on the leader–follower method, the virtual leader was used to determine the desired position of USVs in formation, and the streamlines generated by the EIFDS guided the USVs. In order to make the formation converge to the desired formation better, the vector and scalar of the EIFDS algorithm were uncoupled, and different designs were made to achieve convergence to the desired formation. The interfered residue of the IFDS is not suitable for addressing collision avoidance between USVs in practice. Therefore, the vector field method was employed to tackle the issue, with some enhancements made to optimize its performance. Subsequently, a weighted separation method was applied to combine the vector field and EIFDS, resulting in a composite field solution. Finally, the formation obstacle avoidance strategy based on composite fields was formed. The feasibility of this scheme was verified by simulation, and compared with the single IFDS formation method, the pairwise spacing of USVs behind obstacles could be increased, and the reliability of formation obstacle avoidance was increased.

The marine environment is highly complex and variable, featuring obstacles such as islands, buoys, and vessels. Safe navigation of the surface agent (SA) fleet is crucial for ensuring the safety of the SA fleet, enhancing operational efficiency, and guaranteeing the smooth execution of the fleet’s mission. Regarding the problem of formation obstacle avoidance for SA fleets encountering complex obstacles during navigation, this chapter presents a fixed-time-based safe navigation algorithm for the SA fleet based on streamline traction. Firstly, to precisely position each SA at the designated location within the formation, a highly malleable leader–follower formation mode is introduced. Based on an enhanced interfered fluid dynamical system (EIFDS) obstacle avoidance algorithm, the virtual Leader is guided to evade static obstacles and determine a trajectory of the designated position. Secondly, a first-order fixed-time control Lyapunov function (FTCLF) is designed based on the EIFDS obstacle avoidance algorithm to guide the angular velocity constraint. The optimal guiding angular velocity signal is obtained through quadratic programming, ensuring that the SA steers towards the designated position while avoiding obstacles. Next, for the guiding velocity amplitude signal, a first-order fixed-time control barrier function (FTCBF) is designed based on the streamline formation scheme and the inter-boat safety distance to guide the velocity amplitude constraint. The optimal guiding velocity amplitude signal is obtained through quadratic programming, guaranteeing that each SA maintains the formation while avoiding collisions with adjacent vessels. Finally, the simulation results indicate the effectiveness, superiority, and stability of the proposed fixed-time-based safe navigation guidance algorithm for the SA fleet based on streamline traction.

Communication radiation source individual identification technology is an essential technique in electronic reconnaissance and a crucial link in electronic warfare support measures. Nevertheless, when the sample set encounters complex circumstances, such as class imbalance or a small sample, the classification network model, driven by big data, disrupts the symmetry between the recognition effect and the quantity of the datasets, leading to suboptimal recognition performance. Thus, it is requisite to optimize the existing models and algorithms to better propose more representative fingerprint features. This paper references the speech signal recognition model multivariate long short-term memory–fully convolutional network (MLSTM-FCN), and ameliorates the recognition algorithm and training strategy for the two scenarios of class imbalance and a small sample. It puts forward a communication radiation source individual identification method based on MLSTM-FCN incremental random feature concatenation and a communication radiation source individual identification method based on meta-learning. Proceeding from improving the class imbalance issue among features and small-sample learning, the experimental results under various signal-to-noise ratios demonstrate that the proposed methods have superior recognition effects and higher accuracy.