Towards Autonomous Air Traffic Control for Sequencing and Separation - A Deep Reinforcement Learning Approach

摘要

In this research, we study the feasibility and solution performance of using an artificial intelligent agent to perform the sequencing and separation task in air traffic control. Our goal is to safely separate the air traffic and minimizing the delay. To do this, we have the options to select the most efficient route and provide speed change commands for each aircraft. We use the NASA Sector 33 as our simulator to demonstrate our proposed method. Our results show that our artificial intelligent agent beats most of the human players in this simulation environment and is able to maintain safe separation and sequencing between the aircraft.