Human-like speed modeling for autonomous vehicles during car-following at intersections

This study aims to model drivers’ speed in car-following during braking situations at intersections estimate a safe comfortable human-like the minimum distance for autonomous vehicles (AV). Several behavioral measures were extracted different times before reaching following and intersection control type (signalized or unsignalized) was recorded train using three machine-learning techniques. The results showed that XGBoost is superior other techniques with R 2 values of 0.99 0.97 training testing datasets, respectively. also indicated impacts driver distance. modeled will provide more experience AV riders not violate expectations surrounding traditional vehicle drivers. Also, proposed can be adopted enhance current models by considering effect their type.