Investigation of the contributing factors of driver takeover time under conditional autonomous driving conditions

Investigation of various factors’ influence on driver takeover time is crucial under conditional autonomous driving conditions. It can significantly impact the transition of the control process from driving autonomation to manual control and overall driving safety. In this study, utilizing meta-analysis data, 18 influencing factors were identified and investigated to determine their effects on the takeover time. The random parameter accelerated failure time model, known for its high goodness of fit and competence in accounting for the unobserved heterogeneity, was selected for modelling and analyzing the takeover time. The research focused on understanding how different factors influence the takeover time. The findings reveal that activities such as using handheld devices for tasks, higher levels of automation, performing visual tasks as nondriving related tasks (NDRTs), engaging in high cognitive load tasks, responding to visual takeover requests, and encountering interactive factors with other road users all contribute to longer takeover times. Conversely, NDRTs involving visual methods, manual operations for tasks, the existence of takeover requests, tactile takeover requests, high-fidelity simulators, urgent takeover requests, and complex driver reactions were found to shorten takeover time. In conclusion, four suggestions are proposed to optimize the safety design of the autonomous driving system, considering both the takeover time and takeover quality, thereby providing valuable theoretical references for enhancing driving safety.