Dynamic Non-Binary Prioritisation for UTM Resource Allocation

Abstract

Advanced air mobility (AAM) is set to revolutionise aerial operations, with uncrewed aircraft system (UAS) missions varying significantly in societal importance, urgency, and scheduling requirements. Existing regulatory frameworks, however, rely on static and binary prioritisation schemes which fail to address the complexity and diversity of UAS missions. This approach is particularly problematic in federated UAS traffic management (UTM) environments, where no single actor allocates interdependent resources. Notably, such environments often default to a first-come, first-served (FCFS) approach, leading to inefficiencies, delays, and suboptimal scheduling outcomes. This study thereby proposes a dynamic and non-binary prioritisation scheme to address the limitations of conventional static approaches. It defines the requirements and objectives of such a scheme, and contextualises the framework within a federated UTM architecture aligned with global concepts of operations (ConOps). Notably, extensive simulations confirm that non-binary priorities can effectively improve the success rate of socially critical missions in a federated UTM ecosystem. Moreover, focus groups with UTM stakeholders are leveraged to identify generic characteristics that may influence mission priorities, and expert-defined prioritisation functions are derived using a bespoke symbolic regression tool. Supervised learning through a Siamese network is further used to learn priority mappings from pairwise expert decisions. This data-driven technique is identified as a more practical approach to priority mapping, and shown to outperform expert-defined functions even in simple environments.