Using Agents to Evaluate Performance and Ambiguity of Decentralized Advanced Air Mobility Operating Practices

Advanced Air Mobility (AAM) utilizes electric Vertical Takeoff and Landing (eVTOL) aircraft to transport passengers and cargo. The Federal Aviation Administration (FAA) is exploring shifting strategic deconfliction, demand and capacity balancing, conformance monitoring, and other traffic management responsibilities for these future AAM operations to private third parties under the Extensible Traffic Management (xTM) concept. Stakeholder-developed agreements between operators and the FAA, known as Cooperative Operating Practices (COPs), will be part of the xTM concept. This paper presents an agent-based modeling approach to evaluate decentralized strategic deconfliction COPs, which were used to schedule AAM flights based on operational priority rules without a centralized service provider or authority. COPs were found to be fair from the AAM operator perspective based on simulated fixed and on-demand scheduling within a seven-vertiport network in the Los Angeles area. Schedule stability was most impacted by aircraft departing at a different time than scheduled. An ambiguity analysis, leveraging a large language model (LLM), evaluated whether scenarios arise in which the operational priority is difficult to discern based on the COPs. This type of analysis can inform revisions to COPs that reduce inconsistencies during scheduling. Explicitly assigning priorities to landed aircraft and providing tiebreakers that apply under all conditions were necessary revisions identified by the ambiguity analysis.