AI in the sky: The A2A3 testbed

Future air traffic control systems will detect conflicts up to 20 minutes or more into the future and automatically generate solutions to the conflicts detected. In support of this effort, a testbed has been created utilizing embedded AI technology. The testbed consists of components that maintain detailed models of aircraft trajectories, detect conflicts with those trajectories, generate maneuvers to resolve the conflicts, and provide a graphical interface generation component which uses heuristic search and continuous constraint satisfaction techniques implemented with a knowledge-based system tool. The testbed was implemented on a RISC System/6000. It consists of the automated problem resolution (APR) component, and an architecture to support the development and analysis of APR algorithms and their performance. These issues are discussed in detail. APR will generate maneuvers to resolve conflicts by adjusting aircraft trajectories in each dimension using a specific set of generic maneuver types