In the study, the authors observed 26 men and women (ages 20-52) engaged in simulated driving at 70 mph with and without a non-driving secondary (i.e., distracting) task and recorded the response time as the drivers took over or relinquished control of the automated system.

A takeover request was issued at random intervals ranging from 30 to 45 seconds during normal motorway-driving conditions. The authors found that drivers engaged in a secondary task prior to a control transition took longer to respond, posing a safety hazard. The control transition times were substantially longer than those reported in the peer-reviewed literature.

Boxplot of the takeover reaction time (TOrt) when switching from automated to manual control in the two experimental conditions (no secondary task and secondary task) contrasted with the TOrt of reviewed papers. Eriksson and Stanton (2017) Click to enlarge.

The authors note that their findings can guide policy makers in setting guidelines for how much lead time a driver will need before taking over from the automation.

In this study we found that the range of time in which drivers resume control from the automated driving system was between 1.9 and 25.7 s depending on task engagement. The spread of TOrt in the two conditions in this study indicates that mean or median values do not tell the entire story when it comes to control transitions. Notably, the distribution of TOrt approaches platykurtic when drivers are engaged in a secondary task. This finding implies that vehicle manufacturers must adapt to the circumstances, providing more time to drivers who are engaged in secondary tasks while in HAD mode to avoid excluding drivers at the tail of the distribution.

In light of this consideration, designers of automated vehicles should not focus on the mean or median driver when it comes to control transition times. Rather, they should strive to include the larger range of control transitions times so they do not exclude users that fall outside the mean or median. Moreover, policy makers should strive to accommodate these inter- and intra-individual differences in their guidelines for “sufficiently comfortable transition times.” When drivers were allowed to self-regulate the control transition process, few differences could be found in both driving performance and workload between the two conditions. This finding lends further support to the argument for designing for the range of transition times rather than the mean or median in noncritical situations.

Last, based on the large decrease in TOrt kurtosis when drivers were engaged in a secondary task, it may also be the case that future automated vehicles need to adapt the TORlt [takeover-request lead time] to account for drivers engaged in other, non-driving tasks and even adapt TORlt to accommodate external factors, such as traffic density and weather.

—Eriksson and Stanton (2017)

Resources

• Alexander Eriksson, Neville A. Stanton (2017) “Takeover Time in Highly Automated Vehicles: Noncritical Transitions to and From Manual Control” Human Factors doi: 10.1177/0018720816685832