Collision warning systems represent a promising means to reduce rear-end crash involvement. However, these systems experience failures in the real-world that may promote driver distrust and diminish drivers' willingness to comply with warnings. Recent research suggests that not all false alarms (FAs) are detrimental to drivers. However, very few studies have examined how different alarms influence different driving populations.
The purpose of this research was to examine how younger, middle-aged, and older drivers (with and without UFOV impairments) evaluated and responded to four different alarm contexts - false alarm (FA), nuisance alarm (NA), unnecessary alarm (UA) and true alarm (TA) - when they did and did not receive warnings. FA contexts represent out-of-path conflict scenarios where it is difficult for the driver to identify the source of the alarm. NA contexts represent out-of-path conflict scenarios that occur in a predictable manner that allows drivers to identify the source of the alarm. UA contexts are transitioning host conflict scenarios where the system issues an alert but the situation resolves itself before the driver needs to intervene. TA contexts represent in-host conflict scenarios where the situation requires the driver to intervene to avoid a collision.
The results suggest that alarm context does matter. Compared to response data that differentiates FA and NA from UA and TA, subjective data shows greater sensitivity and differentiates between all four alarm contexts (FA
Younger drivers indicated a high degree of confidence in their own ability across the different conditions. While they adopted a similar response pattern as middle-aged drivers during the TA contexts, these drivers responded less frequently than middle-aged and older drivers during the UA context. Diminished hazard perception ability and the tendency to consider these situations less hazardous likely account for the fewer responses made during these situations by younger drivers.
Older drivers with and without UFOV impairments indicated similar hazard ratings for UA and TA contexts, yet drivers with UFOV impairments responded less frequently in both alarm contexts. Diminished hazard perception ability, slower simple response times, and degraded contrast sensitivity likely account for the fewer and slower responses. Interestingly older drivers with impairments did respond more frequently when warned during the TA context. They also rated FAs and NAs more positively than the other driver groups.
The results of this study suggest applying signal detection theory without concern for the alarm context and driver characteristics is insufficient for understanding how different alarms influence operators and that subjective data can inform design. Researchers are encouraged to combine multiple perspectives that incorporate of both an engineering and human perspective.