Explicit Contingency Planning For Improved Human-Autonomy Teaming In Decision Support

Abstract

As the National Airspace System (NAS) evolves into a more automated system, it will be essential that human operators can effectively team with their automated Decision Support Systems (DSSs) to manage the performance of the system. When automated systems recommend courses of action, the human operator must understand the operational recommendations with sufficient depth and clarity to evaluate their appropriateness and monitor the performance of the system. Significant shortcomings exist in the current state-of-the-art in Air Traffic Management (ATM) DSSs that cause human specialists to distrust the automation’s recommendations and information provided by the system.The focus of the research effort described herein is to identify methods, algorithms, and an overall framework in which ATM DSSs can reason about the appropriate contingency plans to consider in different operational scenarios and communicate the contingency plan to the human specialists to fulfill their information needs. This effort also studied approaches to automatically predict the effectiveness of contingency plans, so that the ATM DSS can determine when a given contingency is no longer the best option and a new ‘plan B’ should be considered.