Perceiving Accident Risks of Urban Naturalistic Driving Environments

In previous studies, the characterization indicators of driving environments have not been sufficiently investigated. Based on a case-crossover framework and online ride-hailing accident samples from Shanghai, Shenzhen, and Chongqing, China, the accident risks of a naturalistic driving environment, including road attributes, points of interest (POIs), and elements in the sight view of a driver, were quantitatively perceived. Results indicated that the five types of POIs examined in this study could significantly increase the risk of ride-hailing accidents. Among these, the influence of bus stations was the highest. In terms of the road class, the accident risk on residential roads was the highest. Among the SVI elements, guardrails were found to be hazards, and the background elements in an SVI seemed to have a nonlinear impact on car accident risk. The paper is useful in the elements’ optimization of the urban naturalistic driving environment.