Reduced autonomy workspace (RAW)—an interaction design approach for human-automation cooperation

Abstract

AbstractLack of support for handling a reduction of autonomy in a highly autonomous automation may lead to a stressful situation for a human when forced to take over. We present a design approach, the Reduced Autonomy Workspace, to address this. The starting point is that the human and the automation work together in parallel control processes, but at different levels of autonomy cognitive control, such as setting goals or implementing plans, which is different from levels of automation. When autonomy is reduced, the automation should consult the human by providing information that has been aligned to the level at which the human is working, and the timing of the provision should be adapted to suit the human’s work situation. This is made possible by allowing the automation to monitor the human in a separate process. The combination of these processes, information level alignment and timing of the presentation, are the key characteristics of the Reduced Autonomy Workspace. The Reduced Autonomy Workspace consists of four phases: Identification of the need; evaluation of whether, and, if so, when, and how to present information; perception and response by the human; implementation of a solution by the automation. The timing of the information presentation should be adapted in real-time to provide flexibility, while the level of the information provided should be tuned offline and kept constant to provide predictability. Use of the Reduced Autonomy Workspace can reduce the risk for surprising, stressful hand-over situations, and the need to monitor the automation to avoid them.