Prediction of Vehicle Instantaneous Speed in the Car-Following Based on Machine Learning Approaches

The instantaneous speed prediction plays a crucial role in autonomous driving, which directly affects the safety of the autonomous vehicle. It is necessary to study instantaneous speed prediction approaches in the car-following. In this study, different machine learning approaches are used to predict the instantaneous speed in the car-following (i.e., support vector regression, random forest, and XGBoost and AdaBoost regression models). And then different model evaluation criteria are selected to assess the model’s prediction power, including mean absolute error, mean absolute percentage error, root mean square error, and variance of absolute percentage error. The denoising trajectory data of the next generation simulation (NGSIM) project is used, and the grey relational analysis is used to extract the feature variables. The results indicate that XGBoost model can effectively improve the accuracy of instantaneous speed prediction in the car-following.