The Future of HLA in an Uncrewed Systems Era?
Military simulation has long used HLA to connect simulators for training exercises that are episodic. People rehearse a scenario, then step away, and some of what they learn fades over time, and training runs to a schedule with a defined end. Uncrewed systems do not share that limit. Once deployed, an AI model can keep learning without pause, its behaviour persistent rather than episodic, changing between missions as it is retrained in response to adversaries, new capabilities or revised doctrine, and in some cases even while a mission is underway. Federation agreements and object models, including the Real-time Platform Reference FOM (RPR-FOM) widely used for this kind of exercise, typically assume that participants and their interactions can be defined in advance, for the length of a single exercise, an assumption that systems capable of continuous, persistent learning are only beginning to test. HLA works by having every participant agree in advance what it is and how it will interact, then holding to that agreement for the length of the exercise; an uncrewed system whose behaviour keeps changing unsettles that starting point, since it isn't clear whether an earlier agreement still holds once a system behaves differently from one mission to the next, or during one. The paper draws on joint work for SISO, conference and writing activity on uncrewed systems for the Royal Aeronautical Society, work with serving military pilots and developers of Collaborative Combat Aircraft-type platforms, NATO ET-058 research on modelling and simulation for uncrewed systems, and years of combined engagement with SISO and NATO M&S activity. Collaborative Combat Aircraft, which pair a human pilot with an autonomous wingman, offer the main test case, though the same question, how far a fixed federation model can represent a partnership whose division of labour keeps shifting, applies wherever humans and autonomous systems work together. There is a further aspect: simulation may be starting to author the behaviour that hardware, and the standards describing it, then have to catch up with, rather than the reverse. Current federation practice, HLA included, may fall short of supporting that. The paper considers whether HLA-based approaches can adapt to this challenge, and what common standards and evidence of interoperability would be needed. It asks the ITEC community to address these questions now, while there is time to shape the answer, not after uncrewed systems have been widely fielded and become ubiquitous.