Intelligent interfaces to aircraft systems

Intelligent interfaces can aid the pilot by improving the robustness, speed, and uniformity of the pilot-vehicle interface, b assisting in the operation of aircraft systems, and by adeptkg the pilot mteractions to the overall situation and the pilot's attention and workload. In order to perform these functions, intelligent interfaces must have models of the aircraft systems (including how to use them), models of situations (including how situations influence the pilot), and models of pilot information processing (to provide the basis for cooperation with the pilot). By combining artificial intelligence concepts with the human information processing model of Rasmussen, we have developed an abstraction hierarchy of states of knowled e, processing functions, and shortcuts, which is useful for characterizing both ilot information processing and the ways in which intelfgent systems interact with the pilot. One promising result from this work has been the tentative identification of a particular class of information processing shortcuts, from situation characterizations to ap ropriate responses, as the most robust safe human inf%nation processing athway for dealing with corn lex time critical situations. gituation-response models can& acqusd from specialists, such as test pilots and systems engmeers, and encoded in pilot aiding systems. There is promise that this ap roach to aiding wlll maintain the appropriate level of p&t situational awareness, while mamtaining the peak cognitive workloads at levels more characteristic of situation recognition than of problem solvin We are using this approach in the conceptual design o f k e e kinds of intelligent aiding systems: an integration-level pilot interface management system, a eneral situation-response pilot aiding system, and intelfigent interface systems associated with the aircraft systems.