The effect of two-stage warning system on human performance along with different takeover strategies

Abstract

Warning stages and takeover strategies are two key factors that affect the takeover process when highly automated systems fail. However, the interaction between these two factors remains unclear. This study empirically investigated the effect of warning types (two-stage warning vs. single-stage warning) and takeover strategies (interruption vs. multitasking) on human performance. Thirty-eight participants performed three tasks of the Multi-Attribute Task Battery (MATB), in which the system monitoring task was automated and adopted as the takeover task, while tracking and resource management tasks were employed as the initial tasks simultaneously. The results showed that compared with the single-stage warning system, the two-stage warning system enhanced situation awareness, reduced mental workload and improved takeover performance. However, tracking task performance declined when the two-stage warning system was used. Moreover, participants using the two-stage warning system reacted to gauge malfunctions much more quickly than participants using the single-stage warning system when the interruption strategy was adopted. However, there was no significant difference in tracking task performance between the two warning systems when the multitasking strategy was adopted. This study concluded that the two-stage warning system can benefit the takeover process but at the cost of the performance of initial tasks, and the takeover strategy plays an important role in realizing the benefits of the two-stage warning system and alleviating its costs. These findings could provide useful insights for automated system design and personnel training.