Design and Experimental Validation of Transparent Behavior for a Workload-Adaptive Cognitive Agent

This work describes and validates a concept of transparent behavior for adaptive automation in the field of military helicopter missions. The adaptive automation is implemented as a cognitive agent, to serve as an artificial co-pilot. It dynamically adjusts its level of assistance by choosing from different workload-adapted strategies of assistive intervention. However, adaptive interventions may entail a possible drawback. It might be difficult for the human operator to build up a sufficient and stable mental model of the interaction. For the purpose of creating transparent behavior, this contribution provides an approach for the agent to communicate in a more human-like fashion. To quantify the impacts of the additional transparency information the artificial agent communicated, we conducted a human-in-the-loop experiment. The results revealed an enhancement of situation awareness and an increase of perceivable intelligence and other human-like characteristics of the cognitive agent.