How system accuracy affects the evolution of drivers' trust towards intelligent driving? an eye-tracking study based on simulated driving experiment

Human trust plays a vital role in collaborative control of vehicles between drivers and intelligent driving. This study aimed to explore how system accuracy affects the evolution of drivers’ trust towards intelligent driving systems based on eye-tracking techniques. A mixed experimental design was implemented, where participants were assigned to one of three simulated driving conditions to complete continuous driving tasks assisted by an intelligent driving system. Each of the three driving scenarios was comprised of three trust evolution stages (i.e., establishment, impairment and repair stages), varied by system accuracy levels (i.e., 100%, 75% and 50%) in trust impairment stage. Results showed that trust evolution stage and system accuracy had significant effects on compliance rate, situational trust and eye movement behaviours. Trust evolution stage and system accuracy had significant interaction effects on compliance rate, situational trust. The established human trust was impaired as system accuracy decreased, particularly under low accuracy level, while the impaired trust was repaired when system accuracy recovered to a high level. The findings contribute to the development of dynamic framework for trust evolution in intelligent driving and provide implications for the design of repair strategies for trust impairment in intelligent driving.