Abstract
Fault management (FM) performance and information sampling behavior were examined while participants operated a complex micro-world that simulated system failures within a generic life-support system of a spacecraft. One group of participants was supported by an integrated display that facilitated fault identification while the other group had to rely on separated raw data readings. The effects of two modes of intelligent automation support were also examined. At a low level of automation (LOA), fault identification was generated along with an advisory on a sequence of recovery actions to be implemented manually at the operator's discretion. The higher LOA generated fault identification but allowed the operator a veto against automatic implementation of the suggested recovery plan. Automated fault identification was unreliable 30% percent of the time; in these trials the automation missed fault identification, gave a false alarm to a nominal situation, or provided false or incomplete fault identification. Initiation of proper recovery when the automation failed was less accurate and slowest when operators were supplied with the non-integrated display, but only when they operated at the low LOA mode prior to the failure. This effect was associated with superior information sampling behavior at the higher LOA when automation was reliable. Display integration and the higher LOA thus supported more efficient sampling of fault relevant displays thereby, promoting better fault state awareness in the presence of reliable automation. This also benefited fault management when the automation failed.
Get full access to this article
View all access options for this article.
