Increasingly Autonomous Traffic Flow Management Under Uncertainty

Today, traffic managers largely rely on their intuition for making Traffic Management Initiative (TMI) decisions due to lack of decision aids. As a result TMIs are often inefficient and there is a lot of variability in their application across similar situations. NASA's 'Similar Days in the National Airspace System (NAS)' research addresses this issue, but, the research tools produce not a single recommended TMI choice but an array of choices, with the final decision again left to the manager's intuition. The proposed SBIR research provides a capability for down-selecting to the most effective TMI choice by developing a what-if analysis functionality for exploring multiple TMI options by realistically simulating NAS-wide operations under the influence of individual TMI options. This what-if analysis capability achieves accurate modeling of NAS traffic flows under uncertainty by creatively integrating two innovations. The first is a traffic flow modeling framework for enabling fast and accurate simulation of individual aircraft transits through the NAS network. This traffic flow modeling framework, which we call the Hybrid Traffic Flow model combines desirable features of trajectory-based models with aggregate traffic flow models to allow fast, near real-time NAS performance evaluation under multiple candidate TMI options. Each option is evaluated under multiple scenarios to capture the whole range of possibilities as per the underlying real world uncertainties,. The second is Bayesian Networks for modeling variations caused by underlying NAS uncertainty factors with explicit encoding of human reasoning behind multiple influencing decisions (e.g., Center MIT restriction impositions, airline cancellations), this enables realistic traffic demand and capacity forecasting for feeding the traffic flow model-based TFM evaluations.