Since the beginning of the 21st century, the ocean has assumed growing strategic importance in resource development, environmental monitoring, and national security. As a core technology driving the development of marine intelligence, swarms of maritime unmanned surface vehicles (USVs) have demonstrated significant application value in both military and civilian fields, such as military reconnaissance, environmental monitoring, maritime search and rescue, and resource exploration. With the growing strategic importance of the ocean, the demand for large-scale and complex marine missions has become increasingly urgent. Compared with traditional manned vessels and individual USVs, USV swarms possess unique advantages including wider coverage, higher operational efficiency, and stronger robustness, making them a key enabler for national maritime strategies. However, due to the inherent high dynamics, uncertainty, and communication constraints of the marine environment, achieving high-performance and highly reliable swarm control remains an arduous challenge. To address this challenge, this paper adopted a structured framework comprising four key components: marine environmental challenges, swarm control requirements, evolution of control methods, and future development trends. This approach enabled a systematic analysis of research progress in USV swarm control and aims to provide scientific guidance for future development in this field.

This review systematically summarized the research advances in USV swarm control. The characteristics of the marine environment were elaborated, including environmental disturbances and communication limitations that affect control performance. The advancements in USV technologies in China and abroad were summarized, showing the diverse development trends of platforms. Subsequently, the core control requirements and key technical challenges of USV swarms were analyzed, focusing on dynamic formation, environmental adaptability, cooperative control architecture, and autonomous intelligence. Furthermore, three representative swarm control methods were comprehensively reviewed: trajectory-based, path-based, and target-based guidance control. Each method was further categorized into traditional model-driven and emerging data-driven intelligent paradigms. Traditional model-driven methods rely on precise kinematic and dynamic models and stability theory, while data-driven intelligent methods leverage technologies such as fuzzy logic, neural networks, and reinforcement learning to improve adaptability and robustness. By systematically reviewing representative studies within these two paradigms, this paper clarified the technical evolution of control methods in this field, from model-dependent approaches to data-intelligent paradigms.

Currently, significant progress has been made in USV swarm control methods; however, there are still limitations in terms of generalizability, real-time performance, and engineering applicability. Future research should focus on four key directions: first, developing a unified descriptive framework based on graph signal processing to improve the evaluation and repair mechanisms of swarm task capabilities; second, enhancing the swarm's environmental adaptability through tight integration and coordinated operation of communication, perception, decision-making, and control systems; third, promoting deep integration of heterogeneous swarms and cross-domain collaboration (e.g., among UAVs, USVs, and UUVs) to expand the scope of swarm tasks; fourth, integrating intelligent learning techniques with model predictive control to enhance intelligent decision-making and autonomous control in swarms. Conducting research on these aspects could help to transition USV swarm control from theoretical exploration to practical deployment in complex marine environments, providing solid technical support for the Smart Ocean Strategy.