Predictive Vehicle Driving Model Considering Modeling Errors of Center of Gravity, Azimuth Angle, and Sideslip Angle of
摘要
Model-based control, which has attracted attention in recent years in research on autonomous driving, requires a simple vehicle model that can accurately represent vehicle behavior. In this study, in order to estimate vehicle behavior with high accuracy in driving environments such as snowy roads and unpaved roads, we focus on modeling errors (center of gravity error, azimuth angle error, and sideslip angle error) that occur in a kinematic bicycle model, propose a method to learn and estimate these using a three-layer neural network, and demonstrate its usefulness through simulation experiments.