Levels of automation: effects of individual differences on wheelchair control performance and user acceptance

Abstract

This paper presents an evaluation of levels of automation. It draws on a prototype of an electrically powered wheelchair as an example of an ordinary automation device operated by less trained users with a broad range of abilities. A study is described during which 21 participants drove through a standardised course three times. Each time, another level of automation was active. As criteria, joint human-automation performance and user acceptance were measured. As predictors, fine motor abilities were assessed and task complexity was recorded. The analyses showed that higher levels of automation achieved lower user acceptance ratings. In contrast, higher levels of automation fostered better performance especially when less able participants steered the wheelchair and when the routes were more complex.