Missing mediated interruptions in manual assembly: Critical aspects of breakpoint selection.

The factory of the future aims to make manufacturing more effective and easily customisable, using advanced sensors and communications to support information management. In this paper, we examine how breakpoint selection during interruption management can fail, even when using recommendations for interruption management from existing research. We present an experiment based on prior work where mediated interruptions (i.e. smart interruptions that should interrupt at opportune moments) were missed by participants when sent at one of two pre-defined breakpoints. These breakpoints were selected based on existing research to minimise the cost of interruption, which can involve longer times to complete tasks as well as making errors on tasks. Missing mediated interruptions in this way was unexpected, and the prior study was not configured to measure this effect, which has led to the experiment detailed here. We strive to explore whether there is a risk of missing notifications when mediated interruptions are used, and how this is affected by breakpoint selection. This was investigated through an experiment that uses tasks and environments that simulate a manufacturing assembly facility. The results indicate that the effect exists, i.e. that participants miss significantly more notifications when interrupted at fine breakpoints than when interrupted at coarse breakpoints. An embodied cognition perspective was used for analysis of the tasks to understand the cause of the effect. This analysis shows that an overlap between "action" and "anticipation of action" can account for why participants miss notifications at fine breakpoints. Based on these findings, recommendations were developed for designing interruption systems that minimise the costs (errors and time) imposed by interruptions during assembly tasks in manufacturing.