A Taxonomy for System of Autonomous Systems

Abstract

Artificial Intelligence and machine learning is rapidly equipping modern systems with real-time learning and autonomous decision-making capabilities allowing these systems to decide and act on their own based on observations from their operational environment. These autonomous systems become constituent systems in a system of autonomous systems that provides more capabilities to fulfill the mission needs and decreases operators’ workload. These autonomous systems have varying level of autonomy which not only defines their independent autonomous behaviour but also the resulting capabilities of the larger system of autonomous systems. However, there is a gap in system of systems literature to classify different types and key structural heuristics for a system of autonomous systems based on the level of autonomy of its constituent systems. In this paper, we propose a taxonomy for system of autonomous systems using the Observe Orient Act and Decide (OODA) loop as a fundamental abstraction of an autonomous system and leverage it to identify information exchange between constituent systems of the system of autonomous systems. This taxonomy proposes seven levels of information sharing between autonomous systems based on their level of autonomy. We describe a conceptual application of this taxonomy to a search and rescue mission to realize a system of autonomous systems.