Intent Inference-Based Flight-Deck Human-Automation Mode-Confusion Detection

Abstract

In this paper, a new framework based on intent inference is proposed to detect the flight-deck human-automation mode confusion. Due to the rapid advancement of flight-deck technology, human-automation issues have become a core area of focus in today’s aviation safety. The complexity of the advanced flight deck leads to new safety concerns such as dysfunctional interaction between the human and the automation. To reduce the occurrence of incidents caused by the human-automation issues, it is necessary to detect the undesirable interaction between them in a timely manner. To this end, however, it is required to model the behavior of the human and the automation, as well as the interaction between them, which is a challenging task. In this paper, a hybrid system and a discrete event system are proposed to model the complex behaviors of the automation and the pilot, respectively. An intent inference algorithm is then proposed to infer the intents of the human and the automation using the behavior models and sensor measurements. The human-automation mode confusions are then identified by detecting mismatches between the inferred intents of the human and the automation. The proposed framework is demonstrated with illustrative human-automation mode-confusion examples.