Analysis and Prediction of Drivers' Braking Behavior with Different Experience at Right-Angled Turns

Due to the growing number of private cars in recent years, traffic accidents have become a significant social problem. The braking actions of drivers are considered one of the primary causes. Therefore, this paper examined the effects of driver experience on braking behavior at right-angled turns. This driving experiment used Xsens MVN Animate to collect driver braking data. Half of the subj ects were experienced cab drivers, and the others were novice drivers. Then, we used five basic machine learning classical algorithms for driver classification prediction, and the experiments show that SVM has the best prediction result. In order to improve the prediction, this research is based on the Sparrow Search Algorithm (SSA) and improved SSA (ISSA) to optimize the hyper-parameters of SVM by SSA-SVM and ISSA- SVM models. The experimental results show that the ISSA-SVM model has the best prediction result with an accuracy rate of 94.87%, which provides a feasible method for driver experience prediction. This study can provide safe driving advice for novice drivers in the future and is beneficial to the further development of personalized driving assistance systems.