Investigating safety and liability of autonomous vehicles: Bayesian random parameter ordered probit model analysis
Quan Yuan, Xuecai Xu, Tao Wang, Yuzhi Chen
Abstract
Purpose This study aims to investigate the safety and liability of autonomous vehicles (AVs), and identify the contributing factors quantitatively so as to provide potential insights on safety and liability of AVs. Design/methodology/approach The actual crash data were obtained from California DMV and Sohu websites involved in collisions of AVs from 2015 to 2021 with 210 observations. The Bayesian random parameter ordered probit model was proposed to reflect the safety and liability of AVs, respectively, as well as accommodating the heterogeneity issue simultaneously. Findings The findings show that day, location and crash type were significant factors of injury severity while location and crash reason were significant influencing the liability. Originality/value The results provide meaningful countermeasures to support the policymakers or practitioners making strategies or regulations about AV safety and liability.