Longitudinal Driving Behavior before, during, and after a Left-Turn Movement at Signalized Intersections: A Naturalistic Driving Study in China

A human-like driving model can help to improve the acceptance and safety of automated systems (ADS). To performance interaction with conventional vehicles ADS, speed behavior left-turn at signalized intersection was studied using natural data. In this study, 374 valid data points snippets intersections were extracted three phases introduced based on reaction braking, stopping, accelerating in process. Firstly, a one-way ANOVA used study influence traffic density, light state, type, waiting area position each phase spatial distribution speed. The state density main significant effects. Furthermore, analyze acceleration, method frequency contour conducted. butterfly-shaped suggested that “the closer stop line, higher variation acceleration”. Finally, parameters further analyzed. results indicate following: (1) red will lead larger maximum deceleration, starting acceleration. (2) On condition dense more stops duration stop–go may cause time pressure, driver tends choose greater (3) leads distance all phases, whilst increased only increases stop. (4) Both an earlier time. findings provide basis for design assistance vehicles.