Survey on intelligent perception, decision-making and swarm control for maritime unmanned systems

With the expansion of marine observation, resource exploration, intelligent shipping, emergency rescue, and maritime security missions toward deep-sea, long-endurance, multi-platform, and cross-domain operations, maritime unmanned systems are evolving from single-platform autonomy to complex intelligent systems integrating perception, decision-making, and swarm control. This paper systematically reviews their connotation, composition, and technical characteristics, and analyzes the key challenges in complex marine environments. The analysis shows that intelligent perception, decision-making, path planning and navigation, motion control and trajectory tracking, and swarm control jointly form a closed-loop process from environmental cognition and mission decision-making to cooperative execution and trustworthy operation. The main bottlenecks include spatiotemporally consistent representation of cross-domain heterogeneous perception data, autonomous decision-making under uncertain situations and dynamic task reconstruction, swarm control under heterogeneous coupling constraints, and trustworthy operation of purely unmanned autonomous systems in complex missions. Representative research progress and typical applications are summarized, covering marine environmental monitoring and resource exploration, maritime search and rescue, port and infrastructure inspection, and maritime security and military applications. Future trends are further discussed, including acoustic-optical-electric-field multisource perception fusion, game-enhanced intelligent decision-making, heterogeneous coupled swarm control, and trustworthy human-machine collaboration, providing a systematic reference for cross-domain coordination, long-term autonomy, and engineering application of maritime unmanned systems.