Flight deck human-automation issue detection via intent inference

Abstract

In this paper, we propose a new framework to detect flight deck human-automation issues using onboard recorded flight data. As aircraft have become increasingly more sophisticated, today's aviation safety challenges include human-automation issues as a core area of focus, especially as the advanced flight deck has been installed into a significant portion of the operational aircraft fleet. The complexity of the advanced flight deck leads to new safety concerns such as dysfunction of the human-automation interaction. The proposed framework is to detect undesirable human-automation interactions from flight data using hybrid system modeling and intent inference. The complex behaviors of the automation and the pilot are modeled as a hybrid system and a discrete event system, respectively. Human-automation issues are then identified by detecting the inconsistencies between the inferred intents of the automation and the pilot. The framework is tested with an illustrative human-automation issue.