Assessing drivers’ performance when automated driver support systems fail with different levels of automation

Abstract

Adaptive cruise control (ACC) systems and lane keeping (LK) systems are two automated driver support systems that serve as convenience systems by changing the role of the driver from a controller to an automation supervisor. These higher levels of automation might impair drivers’ performance when the automated systems fail. This study examined how drivers’ performance differed when an automated system fails in two different levels of automation; with only LK engaged (lower level automation) and with both ACC and LK engaged (higher level automation). Drivers’ performance in conjunction with automated driving was compared in different road conditions (i.e., straight lane, curve lane, and in traffic). In addition, it was expected that drivers’ characteristics and attitudes towards automation, such as their automation complacency, interpersonal trust, acceptance of the LK, and trust of the LK would also influence driving performance. Results showed that participants had significantly longer response times to a system failure with a higher level of automation than with a lower level of automation. The higher level automation also led to significantly larger maximum lane deviation when a system failed. In addition, participants with higher level of interpersonal trust had significantly larger maximum lane deviation than participants with lower level of interpersonal trust. Thus, these results suggest that failures of higher levels of automation may lead to poorer driving performance during the detection and recovery from the failure and drivers’ level of interpersonal trust affects their detection and recovery from system failures.