Vehicle Fusion Tracking Based on Square Root
Unscented Kalman Filter
Chen Ying,Han Chongzhao
(School of Electronics and Information Engineering,Xi'an Jiaotong University,Xi'an
710049,China)
Abstract:Aiming at the non-linearity of tracking systems and the maneuverability of the
road vehicle movement,a new fusion algorithm based on square root-unscented Kaman
filter(SR-UKF)was presented.A dynamic model,which better satisfied the mobile performance
of the target and fully utilized multi-sensor measurements,was adopted to establish the
state and measurement equations.UKF-based data fusion algorithm was proposed to handle
nonlinear problem of the tracking system,and eliminate the errors caused by linearization
of extended Kalman filter (EKF).Furthermore, covariance square root matrix, instead of
covariance one, was taken in filter iteration, which effectively avoided filtering
divergence,and meanwhile improved the convergence velocity and stability of the
algorithm.The experiments show that the position and orientation tracking accuracy by
SR-UKF-based fusion algorithm are improved 18.22% and 34.81% respectively compared with
EKF-based one.
Keywords:vehicle tracking;nonlinear filter;data fusion