Analysis of Naturalistic Driving Events in Honolulu (Pre-COVID)

ABSTRACT

Dashboard cameras and sensors were installed in 233 taxi vans in Honolulu, Hawaii. They produced many hours of naturalistic driving data (NDD) between fall 2019 and spring 2020 in the form of 20 s recorded events. The study achieved its objectives to (1) collect data from naturalistic driving events where driving maneuvers caused acceleration or deceleration in any direction of 0.5g or higher, (2) develop a database suitable for statistical analysis, (3) derive basic statistics for all variables, and (4) investigate correlations between variables. The database included a total of 402 harsh events, of which were 398 near-crashes and 4 were crashes. Several variables such as road, environmental, driver, and vehicle characteristics were coded for each event. The installation of an NDD system by the taxi company proved to be a successful tool for coaching drivers, and for providing statistically significant insights into traffic safety factors relating to near-miss events, such as wider expressways come with a higher risk for near rear-end events;urban roads without parking lower the risk of near rear-end events;light traffic density significantly reduces the risk of rear-end events on freeways;and, mobile phone usage has a positive and significant coefficient that increases the risk of highway rear-end events.