Manoeuvring and docking of large containerships is often assisted by manually operated tugboats, which is a typically complex and, time-consuming process. The success of this operation is affected by factors including the coordination between the master or pilot of, the assisted vessel and the masters of the tugboats, and the knowledge of a local port conditions and characteristics (e.g. depth variation, currents etc.). The H2020 project MOSES (AutoMated Vessels and Supply Chain Optimisation for Sustainable Short Sea Shipping) is, developing the architecture for an autonomous tugboat swarm that cooperates with an Automated Mooring System for manoeuvring, and docking large containerships. The expected benefits from automating this process include reductions in manoeuvring and docking, times, as well as improvements in operational safety. The objective of this paper is to identify the functional requirements and critical, operational scenarios, as well as to describe the architecture that enables autonomous operation for the tugboat swarm. The, methodology implemented includes the following steps: 1) identifying operational states and autonomy level by exploiting expert, knowledge, 2) identifying functional requirements, and 3) developing the hardware and software architecture. To fulfil the identified, requirements, the autonomous tugboat swarm architecture is designed to be modular and includes the following components: i), detection module, which is responsible for processing real-time data from sensors, ii) path planning module, which includes motion, control and is responsible for navigation and manoeuvring, and iii) control module, which translates the high-level decisions from the, navigation algorithms into actionable steering and engine control commands.