Vehicle Sideslip Angle Estimation Using Finite Memory Estimation and Dynamics/Kinematics Model Fusion Based on Neural Networks
G. H. Lee, Donghyun Kim, Jung Min Pak, Choon Ki Ahn
Abstract
Estimation of the sideslip angle is essential for enhancing the driving stability of an automobile while cornering. This study proposes an algorithm that estimates the sideslip angle by combining dynamics and kinematics models without requiring supplementary sensors. We employ a neural network to choose the weights for the fusion of the two models, which makes the kinematics model fully usable. Moreover, we design a new finite memory estimation algorithm to estimate the sideslip angle using the model, which is robust against modeling errors that can occur in dynamics/kinematics models. The proposed algorithm effectively addresses the deficiencies of the dynamics and kinematics models, affording accurate estimation. The performance of the proposed algorithm is evaluated by simulation using the CarSim program and experiments using data from driving tests of an actual vehicle.