Non-driving-related tasks and drivers' takeover time: A meta-analysis

In the pre-era of fully automated vehicles, humans occupy a pivotal role within the driving system. Extensive research has been conducted to explore how drivers interact with automated vehicles across diverse scenarios. This article presents a meta-analysis of 42 papers to examine the influence of non-driving-related tasks (NDRTs) on takeover time (TOT). To consolidate the effect of NDRTs on TOT, this paper classified the NDRTs into four dimensions (visual, auditory, motoric, and mental), which aimed to provide a unified understanding of this impact. This paper employed the following three analyses to understand this impact. Firstly, a synthetical analysis was conducted to compare the effect sizes across primary studies. Secondly, a two-group analysis was performed on studies that included eligible control and experiment groups. Lastly, a moderator analysis, incorporating seven potential moderators, was conducted to further explore the underlying mechanism. The results from the synthetical and two-group analyses revealed that both visual-mental-motoric (Vi-Me-Mo) and visual-mental (Vi-Me) tasks could increase TOT, with the former having a greater effect than the latter. The moderator analysis, including subgroup analysis and meta-regression, further confirmed the significance of moderators within their respective subgroups. However, most moderators exhibited non-significant effects across different scenarios. The findings of this study underscore the crucial importance of attending to TOT and tailoring automated driving systems based on individual driver characteristics. Furthermore, this paper contributes significantly to the advancement of scientific research and engineering design by providing valuable insights into the automotive industry.