Scott J. Bayles and Bikas K. Das
The Federal Aviation Administration (FAA) is in the process of defining and documenting the operations concept and architecture for the future Traffic Flow Management (TFM) system. The challenge of future TFM is to organize complex air traffic flows through busy areas in the National Airspace System (NAS), manage the volume of traffic into and out congested airport areas, and minimize delay-related problems in the advent of continued growth of air traffic and its complexity. The goal of this project was simply to determine if artificial intelligence (AI) techniques can be applied in a useful way TFM problem resolution. This paper describes a particular class of decision making that relies on past experiences, and how this applies to the TFM domain. We have adopted a casebased reasoning (CBR) approach (Kolodner 1991) recognize "similar" problems and to guide TFM decision making by looking at and reasoning about past situations. This paper describes how the eventual users of such a tool, the TFM specialists, feel about this CBR methodology. Finally, this paper provides a mechanism for documenting the many lessons that we learned over the course of this project.