An Empirically Based Suggestion for Reformulating the Glance Duration Criteria in NHTSA's Visual-Manual Interaction Guidelines

摘要

NHTSA recently proposed performance guidelines for visual-manual interaction with non-driving related in-vehicle systems. While a commendable effort to reduce distraction related crashes, in part they seem overly strict. In particular, NHTSA proposes that for each driver performing a secondary task, no more than 15% of the off-road eye glances can be longer than 2.0 s, and 21 in 24 drivers must meet this criterion. The applicability of this criterion was assessed in a study using data from two eye-tracker based studies, involving 35 subjects performing a range of secondary tasks on normal roads. Results showed that over tasks, the average off-road glance duration lengths were quite robust within drivers but varied widely between drivers. Off-road glance duration length thus seems more to reflect individual driver attention allocation strategy than in-vehicle task complexity. Also, several drivers failed to meet the suggested criterion. Assuming that their relative prevalence can be generalized to the general driver population, then as many as one in six drivers may display the type of naturally long off-road glances that will make them fail to meet the criterion. It follows that any task tested by a group of randomly selected drivers likely will fail, since the suggested performance criterion does not allow for this natural driver variability. As currently written, the proposed compliance testing thus risks disqualifying many in-vehicle systems independently of how well they are designed. The criterion therefore needs to be reformulated, e.g., by measuring compliance on a group level rather than on an individual level.