Run-off-road (ROR) crashes contribute to more than half of the vehicular fatalities nationwide (Fatality Analysis Reporting, 2013). ROR crashes are of particular concern for two-lane rural roadways throughout Minnesota because of their overrepresentation within fatal crashes (Minnesota Crash Facts, 2013). Minimizing the propensity of lane departure events is an ongoing national research goal. Currently, in-vehicle warning systems that alert drivers to an impending ROR event are in the early stages of development and have little consistency in the driver— vehicle interfaces they employ. Understanding how the modality of such systems (i.e., visual, auditory, and haptic) impact driver behaviour will aid in developing appropriate and timely warning systems. Furthermore, assessing driver trust and reliance on such systems will allow us to better understand the driver—computer interaction involved and refine the systems to present the most efficient and effective alerts. Examining driver responses in controlled experimental settings offers invaluable insight to guide the future development of these systems. In the current study, behavioural responses were examined through the use of an in-vehicle haptic-based lane departure warning system (LDWS) in a driving simulator. The study incorporated systematic variation in both the reliability of the system (i.e., the likelihood–100%, 90%, or 70%–that the system was activated for an ROR event) and the sequence of treatment conditions (i.e., the order of inactive system exposure (baseline) and active system exposures (LDWS exposure drives 1 and 2)). The study investigated driver responses to the system in terms of overall system efficacy and the efficacy of the three reliability levels. An additional analysis examined the presence of behavioural adaptation after repeated exposure to the system. Behavioural adaptation is a common secondary effect that occurs when drivers become over-reliant on a safety device (Rudin-Brown and Jamson, 2013). The severity of an ROR event was measured by the total time out of lane (TTL) and maximum lane deviation (MLD). These measures offer the best-simulated representation of the severity of a lane departure event and demonstrate the efficacy of an LDWS to help drivers safely return to their lane. Additional covariates such as road shape (e.g., curved vs. straight), speed, brake pressure, and age were examined to determine the influence they may have on the severity of an ROR event when it occurs. The overall results of the study reveal the effectiveness of the LDWS. TTL was significantly longer when no system was active (baseline) compared to when it was active (LDWS exposure drives 1 and 2). LDWS led to a shorter duration of lane departure events. Numerous covariates acted as predictors to TTL and most were associated with greater velocity, suggesting that if drivers lower their speed, they can return to their lane more quickly when they unexpectedly exit the lane. MLD was also greater for baseline drives compared to LDWS exposure drives. Finally, when participants drove without the LDWS (e.g., baseline) after repeated exposures to it, they maintained significantly reduced deviation measures (i.e., TTL and MLD), suggesting long-term benefits of the LDWS. The covariate of overlapping secondary tasks was found to be a significant predictor for MLD and provided insight into the relationship between distraction and severity of lane departure. Drivers who actively engaged in a distraction task were more likely to travel greater distances when they unexpectedly leave their lane, which can possibly put them at a greater risk of striking a bicyclist, highway worker, or roadside infrastructure (e.g., signage) on the road. Future on-road studies are needed to examine low-cost solutions to in-vehicle warning systems. (Author/publisher)
Samenvatting