Applying Learning Systems Theory to Model Cognitive Unmanned Aerial Vehicles

With the increasing use of machine learning in autonomous systems, there is both an increasing need and interest in models of machine learning that are compatible with models of autonomous systems. Specifically, modern systems engineering processes rely on explicit models of the inputs, outputs, and interfaces of system components, as can be captured using systems modeling languages like SysML. While there is not a consensus standard reference architecture for machine learning, recently, abstract systems theory has been extended with learning theory to form abstract learning systems theory (ALST), thereby providing formal systems models of learning that are compatible with the majority of machine learning solution methods. In this paper, we apply ALST to provide explicit systems models for a cognitive path planning system in an unmanned aerial vehicle (UAV). Importantly, we consider levels of abstraction in the models of the cognitive system. We then discuss how these models enable systems engineering processes. This paper demonstrates an important application of systems theory that brings explicit systems models and model-based systems engineering into cognitive communications and aerospace applications.