In recent years, unmanned surface vehicles (USVs) have gained increasing attention in industry due to their efficiency and versatility in marine operations. Artificial potential field (APF) methods, with their strong adaptability and simplicity of implementation, are widely used in USV path planning tasks. However, the naive APF method struggles in static complex environments, due to the local minima problem. Not to mention that actual navigations may involve other dynamic traffic participants. In this work, an improved APF algorithm integrating the boundary potential field method and the International Regulations for Preventing Collisions at Sea (COLREGs) is proposed. By incorporating the boundary potential field method, this novel approach effectively reduces the computational burden caused by clusters of land obstacles in complex environments, significantly improving computational efficiency. Furthermore, the APF method is refined to ensure the algorithm strictly adheres to COLREGs in head-on, overtaking, and crossing encounters, generating smooth and safe collision avoidance paths. The proposed method was tested in numerous complex scenarios derived from electronic navigational charts. The simulation results demonstrated the robustness and efficiency of the proposed algorithm for collision avoidance within complex maritime environments, providing reliable technical support for autonomous obstacle avoidance in dynamic ocean conditions.

This study delves into the development of an advanced path planning algorithm for autonomous service vessels within offshore wind farms, leveraging the power of Deep Reinforcement Learning (DRL) to surmount the navigational complexities and unpredictable conditions prevalent in marine environments. Central to this innovation is the strategic incorporation of S-57 nautical chart data, meticulously processed to train a modified Deep Q-network (DQN) algorithm. This approach, coupled with an innovative two-channel state design, enables a refined state representation that distinguishes between collision volumes and navigational waypoints with unparalleled clarity, enhances the system's ability to differentiate between critical navigation elements even under overlapping conditions. Together with fine-tuned reward design and implementation of experience pre-processing, the proposed algorithm's performance was evaluated against a conventional baseline model, revealing significant enhancements in learning speed and operational efficiency in both complex unknown marine conditions and the specialized context of offshore wind farms. The findings underscore the algorithm's rapid adaptability and robustness to diverse navigational challenges, and its effective functionality within offshore wind farm settings without requiring explicit scenario-specific training. This research lays a foundation for further exploration into the efficacy and applicability of AI in enhancing maritime navigation and operational decision-making.

To address the challenge that unmanned surface vehicles (USVs) struggle to effectively balance tracking accuracy, control smoothness, and system energy efficiency under external disturbances, this paper proposes an anti-disturbance path tracking control method integrating quantum-inspired optimization (QIO) and dynamic game theory (GT). The proposed control method consists of a two-layer optimization architecture: the upper layer employs dynamic game theory to optimize the guidance process, modeling the optimization of the look-ahead distance ($Ld$) and switching radius ($R$) in the LOS guidance algorithm as a non-cooperative game, and achieves adaptive adjustment to path variations and environmental disturbances by solving for the Nash equilibrium. The lower layer, based on a quantum-inspired optimization algorithm, enhances the control process by employing quantum bit probability amplitude encoding for the PID parameter space and utilizing a quantum rotation gate mechanism for efficient global search, thereby achieving online self-tuning of PID parameters under environmental disturbances. Simulation results indicate that, under sea conditions with external disturbances, the proposed method achieves a superior balance among tracking accuracy, control smoothness, and system energy efficiency compared to the traditional fixed-parameter PID-LOS approach, enhancing the comprehensive anti-disturbance robustness of the USV.

Deep reinforcement learning (DRL) enables model-free adaptive control for marine dynamic positioning (DP) under changing disturbances. However, its application remains limited for vessels with strict mechanical constraints, such as the common stern-only dual-azimuth configuration. This paper applies DSAC-T (Distributional Soft Actor-Critic with Three Refinements), a distributional off-policy deep reinforcement learning algorithm, to DP of an 83 m offshore supply vessel under this stern-only configuration. DSAC-T is compared with proximal policy optimization (PPO) and soft actor–critic (SAC) using a shared reward and observation design and an equal simulation budget. All DRL controllers, together with a fixed-gain proportional–derivative (PD) baseline, are evaluated in simulations covering eight wind directions at two wind speeds, together with a calm-water case. DSAC-T achieves the lowest aggregate position-keeping error among the three learning-based methods with lower propulsor-command activity than PPO. We also find that a tighter Gaussian position tolerance in the reward function substantially reduces steady-state position error with a small increase in propulsor-command activity, identifying this tolerance as an influential design parameter for stern-only DP.