Application of Unscented Kalman Filter to Vehicle State Estimation

Abstract

The unscented Kalman filter (UKF) has become a new technique used in a number of nonlinear estimation application. Its application process includes calculating and transmitting the mean and covariance; making use of the forecast sample points and weighing calculation to forecast the mean and covariance; forecasting measurement value and covariance; at last, calculating the UKF gain, renewing state vector and variance. Some vehicle state variables are not easy to obtain accurately while the vehicle is in motion. However these state variables are of great significance to chassis control. This paper sets up a nonlinear 3 degree-of-freedom vehicle model including yaw motion, longitudinal motion and side motion, and proposes an unscented Kalman filter which generates better estimation of the vehicle state. The accuracy of the estimation algorithm of UKF for estimating yaw rate, and side slip angle, especially the great performance of the estimation of the yaw rate is verified by experimental data of several ISO tests. The result shows that the UKF estimation of vehicle state matches the measured vehicle state very well.