AI Enabling Australia’s Future Submarine
There are significant opportunities to increase the operational capability of Australia’s Future Submarine through the application of Artificial Intelligence techniques to combat system functions. This project assessed these opportunities and outlined a roadmap for the implementation of an Artificial Intelligence capability for Australia’s Future Submarine to support the broader strategic goal of SEA1000 to deliver a regionally superior submarine.
This project aimed to define the problem space for the application of Artificial Intelligence approaches such as Machine Learning (ML) in the submarine context; and to enable the data management system of the Future Submarine (FSM) to support the collection and exploitation of data using these advanced algorithms.
The key deliverable of this project was a problem definition statement for each of the detection, tracking, localisation and classification problems. During the preliminary research for this project it was assessed that the problem space presented more opportunities than those originally envisaged. By examining the problem space through a situational awareness framework, the potential application of the Machine Learning (ML) approaches could be extended to include higher level cognitive functions such as enhancing command decision making and submarine Command Team performance.
The problem space, for the purposes of the DIP project, was limited to that of the Combat System domain and specifically the key functions undertaken by the Command Team.
A total of 13 problems were identified and defined during the analysis. Each problem was defined in the form of a question with the input and output data described. The applicability of ML to the problem was based on the perspective of the University of Adelaide’s Australian Institute of Machine Learning (AIML), while the Defence Science and Technology Group (DST) provided insight into the availability of data that could be used to train the neural networks. Acacia provided operational insight as to which aspects of these problems provide the greatest operational impact.
After a detailed examination of the problem space and having gained insights through collaboration between the key stakeholders, it was concluded that the best approach to ultimately developing the high-level design of FSM data management system was to develop a prototype ML application focused on one of the problems spaces identified in this paper. This is outside the original scope of the DIP project and the key stakeholder partners have agreed to explore this option during subsequent stages of this project.
Acacia Systems Pty Ltd