Position estimation for intelligent vehicles using an unscented Kalman filter

Abstract

This paper proposes a data fusion method for intelligent vehicles, which integrates the data from odometry and magnetic sensors. Errors due to the wheels slippage or roughness of the ground will inevitably be accumulated during the dead-reckoning based on odometry. The magnetic sensors are used to provide absolute positioning reference. The data from odometry and the magnetic sensors are integrated by using an Unscented Kalman Filter (UKF). The UKF method for position estimation is compared with the Extended Kalman Filter (EKF) method. A field experiment is performed, and the results show that the UKF is more accurate in estimating the position of the intelligent vehicle than the EKF.