Research on short-time Prediction of driving behaviour at unsignalized intersections based on machine learning

Abstract

Abstract The unsignalized intersections are the important nodes in the road network, and also the sections with a high incidence of serious collision accidents. Based on the driving simulation experiment, the interactive driving behaviours of multiple drivers were collected. The data were input to the machine learning model, training for predicting the short-time driver’s decision-making. The results showed that the Random Forest model has a better prediction effect than the Support Vector Machine (SVM) model, and the shorter prediction interval also has a better prediction effect. This study was helpful to give timely warning to drivers in danger and provided a new idea for the research of collision prevention warning strategies and control methods.