Resource-bounded reasoning for interface agent for realizing flexible human-machine collaboration

This paper presents the formulation of an interface agent making high-stakes, time-critical decisions. An interface agent's major decision tasks include an autonomous problem solver (i.e., identification and/or diagnosis of system anomalous states and deciding appropriate actions) and a display manager to human users (operators). Attainment of these tasks should be managed under scarce time resources. The formulation is made with respect to the hierarchically organized concepts of plant anomalies that are available to an interface agent by applying a concept formation method.