Rule-based implementation of automation aids for ATC controllers

Abstract

Development of Air Traffic Controller automation aids is frequently hampered by the mathematical nature of the algorithms they are based upon. These limitations are: lack of adaptability to local conditions; high development, testing, and modification costs; and low end-user confidence on the algorithm's behavior. Research in Artificial Intelligence has produced systems whose logic is implemented by means of rules which may be defined by the end-user without explicit programming. Such rule-based systems may provide a flexible, low-cost alternative to mathematical algorithms. Since the end-user can exercise significant control over the behavior of such logic, automation aids built using these methods can be tailored to the user's environment, preferences, and experience more readily than if built around a classical mathematical algorithm. While this is theoretically possible, present rule-based systems technology is insufficient to allow practical implementation of an ATC aid today. An experimental new rule-based core system has been developed which overcomes some of these obstacles, but a number of problems, including that of poor hardware performance, remain outstanding as topic for continuing research.