RL-based scheduling of an AAM traffic network

Abstract

This study presents an approach for pre-flight planning process to be used in the future Advanced Air Mobility (AAM) system especially after contingency situations and relevant activities take place. The methodology for scheduling is modeled as a reinforcement learning (RL) agent that resolves potential conflicts for the traffic and balances the demand and capacity at vertiports. The reason behind to use RL is that specific problem requires a very quick response since it also deals with resolving conflicts that are observed between the flights that are about to take-off and the contingent flights that diverted for an emergency landing. The main objective of this work is to develop a pre-flight planning service to work compatible with contingency management activities for enhancing the contingency management process for the AAM system.