Driver trust, annoyance, and compliance for an automated calendar system

Drivers' interactions with an imperfect automated system were examined while performing a car following task in a driving simulator. It was proposed that primary task workload, alert intensity, and reliability would affect trust, annoyance, and compliance with imperfect secondary task automation. No previous research had examined the dual effects of trust and affect on compliance with an imperfect automated system; therefore, the current experiment was conducted in an attempt to provide a more complete understanding of compliance for an imperfect automated system.;One hundred thirty-five drivers (69 male) participated in the experiment. Drivers completed five experimental trials: a driving-only trial and four dual-task trials where the primary driving task was paired with a secondary automated calendar system. Across dual-task trials, system reliability was manipulated (30%, 50%, 70%, & 90%). The primary car following task was divided into low and high workload conditions, based upon single (lateral) or dual (lateral and headway) tracking perturbations. For the secondary task, drivers interacted with the automated system using probability matching to predict responses based upon system reliability. Finally, alert intensity was varied where drivers either received no tone or a 60 dB, 75 dB, or 90 dB SPL(A) tone designed to signal an upcoming calendar event. Psychophysiological, subjective, and objective performance measures were collected.;Results confirmed a positive relationship between reliability and trust, where trust increased with increasing system reliability. Primary task workload was found to moderate trust, where greater increases in reliability were required to obtain sufficient trust for drivers in the high workload group. Specifically, reliability needed to be 70% for drivers in the low workload group and greater than 70% in the high workload group for trust and compliance to occur. Furthermore, annoyance did not increase with decreasing reliability but did increase as secondary task difficulty increased. Finally, annoyance increased as alert intensity increased; however, it did not affect compliance. Rather, drivers were willing to disable the alert while complying with the automated system.;In conclusion, drivers' trust and compliance towards imperfect secondary automation was predicated on system reliability and primary task workload. Annoyance with the automated system was created by the difficulty of the secondary task and alert intensity; however, compliance was not directly influenced by annoyance in the present example. Instead, the source of the annoyance was not integral to system operation and therefore was deemed expendable through disablement. Overall, these findings imply that drivers' compliance with automated devices is influenced by both ease of use and automation reliability.