Investigation of factors affecting driver performance using adverse condition warning systems

This study addresses issues concerning the design of adverse condition warning systems (ACWS). An ACWS was designed to sense when a car was likely to skid. Through the use of a virtual-driving environment, two important design considerations, alerting system sensitivity (low and high) and type of auditory alarm signal (Binary ON/OFF and Graded) were analyzed and compared with a no-alarm control group. Dependent measures reflecting driver performance and response to the alarm signal were analyzed to understand the effect of different system configurations. Additionally, participants' trust in the alerting system under different configurations was measured to reveal user acceptance in these systems. Participants had fewest skids in the low sensitivity and graded alarm signal condition compared to other alerting system configurations and the no-alarm condition. Participants in the graded signal condition also had a greater degree of lateral control over the vehicle. Additionally, trust was found to be the lower for the high vs. low sensitivity alarm condition, indicating a reduction in trust when the alerting system activated more often (at a lower threshold) perhaps because participants' did not feel the system was accurately reflecting a dangerous condition. Over sessions, positive feelings of trust tended to increase, while negative feelings decreased, as expected given increased experience with the system. This research emphasizes the fact that while ACWS can provide an advantage in terms of vehicle control, characteristics of both the alerting signal, and system configuration must be considered to insure successful outcomes.