Lance Sherry, Director of Mason’s Center for Air Transportation Systems Research in the Volgenau School of Engineering, has two new grants from NASA totaling $250,000. Sherry and his team are looking at a new type of accident scenario. In this type of accident, a perfectly mechanically, structurally, and electronically sound aircraft was flown into the accident. That was nothing broken or malfunctioned, yet the accident happened. The famous researcher Charles Perrow, studying the Three Mile Island Accident, called these accident scenarios "normal accidents" that are the result of tightly coupled complex automation. A small rounding error in a calculation in one automated system can cascade through the overall system leading to an accident. Dr. Sherry and his students have developed a technology called the Paranoid Pilot Associate (PPA) that uses machine learning techniques to process massive amounts of airline operational flight data to predict what kinds of anomalous behaviors might happen in the near future. The PPA, like a "back-set driver" is continuously sifting through the data to warn the pilot what bad outcomes have happened in the past given the current situation. Research continues to minimize the nuisance alerts and run human-in-the-loop experiments with alternate display and aural alerts.
In another NASA-funded project, Sherry and his team are working on developing a safe architecture for a million-fight National Airspace System. The current NAS handles between 30,000 and 60,000 flights a day. Automation that coordinates flight trajectories can avoid collisions, researchers envision. Joined by fellow engineering professors John Shortle, Chien-Chung Huang, Sherry and a team of students will be developing a simulation of a completely automated approach and landing system that could be used at major airports. This autonomous system will adapt to transient and steady-state changes in the environment. It will also demonstrate performance-based risk assessment (as opposed to risk assessment by inspection) to achieve the required target level of safety.
