To effectively fuse the plans of multiple subsystems in a navy ship, we use AI methods to intelligently explore the problem space. We created a module that facilitates the merging of sub-plans with temporal dependencies. This is done by combining all events in the sub-plans, recalculating the temporal dependencies, finding possible incoherence in the final plan and pointing out where conflicts arise. We created a multi-dimensional utility function that is optimised towards achieving the mission goals.

Problem Context

The Royal Netherlands Navy is setting new targets for a higher level of automation for the processes on board future submarines and frigates. Aboard current ships, the crew spend a considerable amount of time on aligning decisions and actions to reach the mission objectives while taking into account the external (hostile) situation, the ship’s status, her systems, the available resources and the availability of the crew. Coordination of actions is mainly performed by operators. Onboard future naval ships, the crew needs a higher level of decision support to cope with the ever-increasing complexity of the systems, systems dependencies, threats and mission goals. Decisions of operators and systems need to be orchestrated to reach the mission goal effectively using the available capabilities. For instance, selecting the optimum propulsion plant configuration with respect to ship signatures and available resources or optimizing navigation with respect to threats in track and weapon assignment. Faster decision and actions in combination with smaller crews require autonomous actions to cope with ever increasingly complex situations. Depending on the task and the workload of the crew, some decisions might be taken autonomously, however, we aim for a decision support system which enables operators to stay on, and in certain moments in, the loop (meaningful human control).

A (fleet of) navy vessel(s) is a complex system of systems operating in a dynamic environment. Each system has its own capabilities and goals while contributing to the global mission. Each subsystem creates plans to realize its own goals based on limited knowledge (world model). Resulting plans of multiple subsystems might require the same resource (e.g. helicopter, radar tracks) resulting in conflicting plans. These conflicts need to be resolved in such a manner that overall performance remains within specified, acceptable bounds.


To investigate how the (potentially conflicting) plans of multiple subsytems can be fused, we define and explore an existing use case. Relating to the Smart Ship project (main stakeholders; DMO, Royal Netherlands Navy), the use case looks at automated frigates. These frigates formulate (sub)plans to guide the vessel's operation and advise the crew. We will explore how to fuse these (sub)plans through several AI techniques (e.g. epistemic logic, dynamic epistemic logic, Q-learning, and heuristic searches such as hill climbing). Each of these techniques aims to intelligently explore solution spaces. We will compare different techniques to find the technique that results in the most optimal solution. The optimal solution can be found by optimising a multi-dimension utility function. The function represents how well the fused plans achieve the various goals of the mission. As a baseline for the comparison, we use a brute force heuristic, which simply tries all possible options and combinations.

Since the mission goals represents both threats to the platform and to the mission, the exploration of the solution space might require specifically directed re-planning for the subsystems (adaptability). The final system will advise the command team in their complex task of commanding the entire vessel while providing explainable insight in the proposed solutions.

Findings are presented in a Conference paper and presentation at the OCEANS 2024 conference.


  • Martijn van den Heuvel, Systems Engineer Autonomous Systems & Robotics, TNO, e-mail: