为了解决无人水面船（USV）在复杂海洋环境中路径跟随的控制问题，本文构建基于MAVLink的通信系统，实现领航船舶与受控船舶间的实时状态传输，确保受控船舶能够根据领航船舶的实时位置、速度等信息进行动态调整，并利用深度Q网络（Deep Q-Network，DQN）的学习方法使受控船舶能够自主学习最优的航行路径，从而提升跟随精度。在通信不稳定的条件下，采用反步控制（Backstepping Control，BC）进行状态预测并实时反馈补偿，从而确保受控船舶能够平稳跟随领航船舶，修正由于数据丢失造成的路径误差。结果表明，该方法在高干扰环境下，尤其在通信延迟和数据包丢失的情况下，仍能维持良好的路径跟随性能。与传统的控制方法相比，基于DQN和BC的混合控制策略显著提高了无人水面船舶的跟随精度和系统稳定性，具有较强的鲁棒性，能够在复杂和动态变化的海洋环境中有效运行。
In order to solve the problem of controlling the path following of unmanned surface vessels (USVs) in complex marine environments, the article constructs a communication system based on MAVLink to realize the real-time state transmission between the pilot vessel and the controlled vessel, and to ensure that the controlled vessel is able to dynamically adjust according to the real-time position, speed and other information of the pilot vessel.And the learning method of Deep Q-Network (DQN) is utilized to enable the controlled ship to learn the optimal sailing path independently, so as to improve the following accuracy.Under unstable communication conditions, Backstepping Control (BC) is used for state prediction and real-time feedback compensation, which ensures that the controlled vessel can follow the pilot vessel smoothly and corrects the path error caused by data loss.The results show that the method can still maintain good path following performance in high interference environments, especially under communication delay and packet loss.Compared with traditional control methods, the hybrid control strategy based on DQN and BC significantly improves the following accuracy and system stability of unmanned surface vessels with strong robustness, and is able to operate effectively in complex and dynamically changing marine environments.
2026,48(3): 145-153 收稿日期：2025-5-13
DOI：10.3404/j.issn.1672-7649.2026.03.023
分类号：U674.91；TP275
基金项目：国家自然科学基金资助项目（62276285）；教育部学位与研究生教育发展中心主题案例库项目（ZT-231028914）；江苏省研究生科研与实践创新计划项目（KXCX25_4373）；中国科学院软件研究所合作项目（2205072325）
作者简介：路春宇(2002-),女,硕士研究生,研究方向为无人船仿真和控制
参考文献：
[1] KAPITANYUK Y A, PROSKURNIKOV A V, CAO M. A guiding vector-field algorithm for path-following control of nonholonomic mobile robots[J]. IEEE Transactions on Control Systems Technology, 2017, 26(4): 1372-1385.
[2] FOSSEN T I, LEKKAS A M. Direct and indirect adaptive integral line‐of‐sight path‐following controllers for marine craft exposed to ocean currents[J]. International Journal of Adaptive Control and Signal Processing, 2017, 31(4): 445-463.
[3] KHAMSEH H B, JANABI-SHARIFI F. Ukf–based lqr control of a manipulating unmanned aerial vehicle[J]. Unmanned Systems, 2017, 5(3): 131-139.
[4] DONG Z, WAN L, LI Y, et al. Trajectory tracking control of underactuated USV based on modified backstepping approach[J]. International Journal of Naval Architecture and Ocean Engineering, 2015, 7(5): 817-832.
[5] XU H, WANG N, ZHAO H, et al. Deep reinforcement learning-based path planning of underactuated surface vessels[J]. Cyber-Physical Systems, 2019, 5(1): 1-17.
[6] DAI J G, GLUZMAN M. Queueing network controls via deep reinforcement learning[J]. Stochastic Systems, 2022, 12(1): 30-67.
[7] MANNUCCI T, VAN KAMPEN E J, DE VISSER C, et al. Safe exploration algorithms for reinforcement learning [controllers[J]. IEEE transactions on neural networks and learning systems, 2017, 29(4): 1069-1081.
[8] MENDA K, CHEN Y C, GRANA J, et al. Deep reinforcement learning for event-driven multi-agent decision processes[J]. IEEE Transactions on Intelligent Transportation Systems, 2018, 20(4): 1259-1268.
[9] MIAO J, SUN X, PENG C, et al. DOPH∞-based path-following control for underactuated marine vehicles with multiple disturbances and constraints[J]. Ocean Engineering, 2022, 266: 113160.
[10] RUBÍ B, MORCEGO B, PÉREZ R. Deep reinforcement learning for quadrotor path following and obstacle avoidance[J]. Deep Learning for Unmanned Systems, 2021: 563-633.
[11] FATHINEZHAD F, DERHAMI V, REZAEIAN M. Supervised fuzzy reinforcement learning for robot navigation[J]. Applied Soft Computing, 2016, 40: 33-41.
[12] LI Z, LIU P, XU C, et al. Reinforcement learning-based variable speed limit control strategy to reduce traffic congestion at freeway recurrent bottlenecks[J]. IEEE Transactions on Intelligent Transportation Systems, 2017, 18(11): 3204-3217.
[13] 郭乃琨, 李修深, 杨继坤. 复杂电磁环境下船舶自动通信系统优化设计研究[J]. 舰船电子工程, 2023, 43(12): 80-85.
GUO N K, LI X S, YANG J K. Research on optimization design of ship automatic communication system under complex electromagnetic environment[J]. Ship Electronic Engineering, 2023, 43(12): 80-85.
[14] 张志柏. 应用于复杂电磁环境下的船舶自动通信系统研究[J]. 舰船科学技术, 2016, 38(6): 139-141.
ZHANG Z B. Research on the automatic communication system of ships in the complex electromagnetic environment[J]. Ship Science and Technology, 2016, 38(6): 139-141.
[15] LI J, CHEN Y, ZHAO X N, et al. An improved DQN path planning algorithm[J]. The Journal of Supercomputing, 2022, 78(1): 616-639.
