Maintaining consistent tracking of surface ships using unmanned surface vehicles (USVs) is particularly challenging due to the target's high maneuverability and complex motion trajectories. Additionally, environmental interferences in marine environments further complicate the task by reducing positioning accuracy. These issues often lead to tracking vibrations and delays, significantly impacting stability and precision. To address these challenges, this study proposes innovative solutions to improve the performance of USVs in maintaining effective target tracking.

This study presents an advanced bidirectional fitting algorithm integrating polynomial fitting and particle swarm optimization (PSO) to address radar sampling errors. By systematically analyzing the correlation between target motion amplitude and radar observation errors, the sampling period is optimized to accurately capture the target's motion patterns. Additionally, the optimal number of sampling points is carefully determined. Polynomial fitting is initially applied to minimize longitudinal errors, followed by secondary horizontal error reduction using PSO. A penalty function is incorporated to impose strict constraints on the fitting range, ensuring that corrected coordinates align with the actual motion capabilities of the target ship. Furthermore, real-time motion data from the USV and the target ship, including target speed, separation distance, and USV performance parameters, are used to develop a robust speed strategy. Geometric methods combined with USV turning dynamics are used to formulate a precise course strategy, enabling optimal speed and trajectory planning.

Rigorous validation through real-vessel experiments shows that the radar data processed with the bidirectional fitting algorithm achieves significantly enhanced smoothness, closely aligning with the ship's actual motion patterns.Moreover, the USV consistently and accurately tracks the target ship by following the optimized speed and course strategies. Vibrations and delays are effectively mitigated, ensuring stable tracking performance throughout the operation process.

The proposed method effectively addresses the challenges of stable target tracking for USVs, demonstrating exceptional performance in single-target scenarios. This work provides essential technical support and practical insights for advancing USV target tracking technology. Its findings provide reference value for further research and offer actionable guidance for broader applications in marine robotics and autonomous systems.

To address the issue of multicollinearity and parameter drift in the identification of hydrodynamic coefficients in ship separated-type models, this paper proposes a method for modeling simplified three-degree-of-freedom modular models based on support vector regression (SVR).

Initially, a processing strategy is introduced to enhance the effectiveness of the sample data. Further, Lasso regression is introduced to select the most influential hydrodynamic coefficients and alleviate multicollinearity. Subsequently, a regression model for the identification of hydrodynamic derivatives is derived for the MMG model. A data centralization and differencing method is then employed to reconstruct the regression model, mitigating the impact of parameter drift on hydrodynamic derivative identification errors.

Simulation experiments demonstrate good agreement between the hydrodynamic coefficient forecast values and numerical simulation results. The calculated values of root mean square error (RMSE) and correlation coefficient (CC) fall within a favorable range.

The SVR algorithm successfully identifies the hydrodynamic derivatives of the modular model, the identified hydrodynamic coefficients exhibit high accuracy, and the established model demonstrates good predictive capability and robustness.

In order to tackle the issue of the poor navigation stability of unmanned surface vehicles (USVs) under interference conditions, an intelligent control parameter adjustment strategy based on the deep reinforcement learning (DRL) method is proposed.

A dynamic model of a USV combining the line-of-sight (LOS) method and PID navigation controller is established to conduct its navigation control tasks. In view of the time-varying characteristics of PID parameters for course control under interference conditions, the DRL theory is introduced. The environmental state, action and reward functions of the intelligent agent are designed to adjust the PID parameters online. An improved deep deterministic policy gradient (DDPG) algorithm is proposed to increase the convergence speed and address the issue of the occurrence of local optima during the training process. Specifically, the original experience pool is separated into success and failure experience pools, and an adaptive sampling mechanism is designed to optimize the experience pool playback structure.

The simulation results show that the improved algorithm converges rapidly with a slightly improved average return in the later stages of training. Under interference conditions, the lateral errors and heading angle deviations of the controller based on the improved DDPG algorithm are reduced significantly. Path tracking can be maintained more steadily after fitting the desired path faster.

The improved algorithm greatly reduces the cost of training time, enhances the steady-state performance of the agent in the later stages of training and achieves more accurate path tracking.

Aiming at the encirclement tactics adopted by enemy ships, this study focuses on the problem of planning an escape strategy when an unmanned surface vehicle (USV) is surrounded by enemy ships.

A hybrid sampling deep Q-network (HS-DQN) reinforcement learning algorithm is proposed which gradually increases the playback frequency of important samples and retains a certain level of exploration to prevent it from falling into local optimization. The state space, action space and reward function are designed to obtain the USV's optimal escape strategy, and its performance is compared with that of the deep Q-network (DQN) algorithm in terms of reward and escape success rate.

The simulation results show that using the HS-DQN algorithm for training increases the escape success rate by 2% and the convergence speed by 20%.

The HS-DQN algorithm can reduce the number of useless explorations and speed up the convergence of the algorithm. The simulation results verify the effectiveness of the USV escape strategy.

A multiple unmanned surface vehicle (USV) cooperative hunting algorithm based on the double layer switching strategy is proposed to cope with the difficulties of USVs in hunting intelligent escaped targets.

Specifically, the first hunting strategy adopts the improved potential point method. The Hungarian algorithm is employed in order to dynamically allocate potential points for USVs, and the optimization goal is applied to minimize the total linear distance between USVs and potential points. In this process, the artificial potential field method is used to achieve cooperative collision avoidance. The second hunting strategy takes advantage of the nature of the Apollonius circle to tighten the surrounding area, i.e. two USVs go to the target point of the escaped target to intercept it, while the remaining USVs maintain the same direction as the escaped target. Moreover, in order to deal with the different escape strategies of targets, the first and second layers of hunting strategy can be transformed into each other.

Numerical simulation shows that the proposed algorithm can reduce the hunting time to less than or equal to that of the sequential distribution potential point algorithm and polar angle distribution potential point algorithm.

The results of this study prove the effectiveness and progressiveness of the proposed algorithm.