Deep reinforcement learning (DRL) has been increasingly used for collision avoidance in unmanned surface vehicles (USVs). However, policies trained under dynamic and disturbed conditions often produce rapid and high amplitude control variations when responding to collision threats in practical multi-vessel encounter scenarios. Such aggressive action changes can accelerate actuator wear and shorten the service life of the control system. To address this issue, a smooth collision-avoidance decision-making method is developed by integrating Mamba sequence modeling, masked attention, and DRL. Firstly, a composite reward function is constructed by jointly accounting for the International Regulations for Preventing Collisions at Sea, velocity-obstacle (VO)-based guidance, navigational safety, and action smoothness, which alleviates reward sparsity and balances safety with control stability. Secondly, Mamba sequence modeling and masked attention are embedded into the actor–critic framework to enable dynamic feature extraction and temporal dependency modeling in multi-target encounters, thereby improving robustness and adaptability in disturbed and uncertain environments. Extensive simulations were conducted against multiple DRL baselines. Under the strongest disturbance with 10 target ships, the proposed method achieved a success rate of 0.87, yielding a 20% improvement over SAC, while the mean per-step torque variation was reduced by 36.5% compared with SAC. In addition, real-field experiments were carried out to validate the learned policy, demonstrating its practical applicability and deployment potential in realistic marine environments.

In complex and dynamic maritime environments, the uncertain navigation intent of target ships (TS) poses a significant challenge for the collision avoidance strategies of unmanned surface vehicles (USVs). To address above issue, an autonomous collision avoidance method that integrates recursive Bayesian estimation, a masked attention mechanism, and the soft actor-critic (SAC) algorithm is proposed in this paper. Specifically, the recursive Bayesian model is employed to estimate the navigation intentions of TS in real time, for enabling the construction of intent-aware state representations that enhance the modeling capability for uncertain target behaviors. In addition, a maximum target count is predefined, and the masked attention mechanism is incorporated into the actor and critic networks of SAC to resolve the mismatch between the fixed input dimensionality of neural networks and the dynamically varying number of TS. Simulation demonstrates that the proposed method outperforms traditional collision avoidance methods and other state-of-the-art deep reinforcement learning (DRL)-based strategies in terms of key performance metrics such as collision avoidance success rate, minimum distance, and average risk. Furthermore, real-field experiments validate the proposed method’s adaptability to real-world maritime environments and its potential for practical deployment.

Scattering properties of lunar surface rocks are important guides for remote sensing observation of the lunar surface and discrimination of geologic features. Previous studies primarily focused on the scattering effects of surface or partially buried rocks in terms of angle or frequency, with limited research addressing SAR simulation images of rocks on the lunar surface. This study proposes a three-dimensional (3D) scattering model that integrates the vector radiative transfer (VRT) method with lunar surface rock scattering mechanisms while effectively combining topographic data, rock abundance data, and radar parameters. To implement this model, an end-to-end computational framework is developed to process input datasets and apply the 3D scattering model for backscattering coefficient simulation, reducing uncertainties in the input parameters. Several case studies are performed on Mini-RF observational data. Simulation results demonstrate that the proposed approach shows good consistency with radar data, particularly in terms of distribution characteristics and average backscattering coefficient errors, with an error margin of less than 3 dB.

The synthetic aperture radar (SAR) imaging environment, especially with limited samples, poses a serious challenge to automatic target recognition (ATR) in modern electronic reconnaissance systems. To enhance SAR image recognition performance, this study proposes a technique leveraging a diffusion model and scattering information. This method involves SAR image generation and scattering information processing to improve generalization with few-shot samples. First, the number of few-shot SAR samples was augmented using the denoising diffusion probabilistic model (DDPM). Then, the scattering information is extracted to stably calculate SAR image similarity from the scattering mechanism. Finally, the recognition task is effectively accomplished through the optimal integration of the recognition network and scattering similarity. Simulation results demonstrate that the proposed method achieves superior SAR image generation quality and recognition accuracy compared to the existing methods when the available data are extremely limited.

A fast and accurate numerical method based on the theory of characteristic mode (TCM) is proposed for analyzing electromagnetic scattering from repetitive multiscale array structures. In order to overcome the low frequency breakdown problem when interacting at short distances between observation and source basis functions, the CMs are extracted from the impedance matrix generated by the augmented electric field integral equation (AEFIE), which is discretized by the method of moments (MoM). Then the CMs of a single unit are used as the global basis functions to reduce the unknowns of the array structures. In addition, the broadband multilevel fast multipole algorithm (MLFMA) based on approximate diagonalization of the Green’s function is implemented to accelerate the matrix-vector multiplications between the impedance matrix and CM vectors. Several numerical examples are given to demonstrate the high efficiency and accuracy of the proposed scheme.

We present a new multilevel fast Fourier transform (MLFFT) method and its application for the 3-D structures capacitance extraction. The multilevel octree structure, multilevel interpolation/projection, and subdomain FFTs techniques are employed in the MLFFT. Distinctions between the MLFFT and the conventional FFT-based methods are discussed carefully. Numerical results demonstrate that the MLFFT is more efficient than conventional FFT based methods, especially when analyzing inhomogeneous problems.