Simultaneous localization and mapping based on iterated square root cubature Kalman filter

Abstract:

In large-scale conditions, the large nonlinear error of simultaneous localization and mapping (SLAM) based on square root cubature Kalman filter (SRCKF) is a serious constraint to high positional accuracy. To solve this problem, an improved SLAM algorithm based on iterated square root cubature Kalman filter (ISRCKF) is proposed. Utilizing the iteration theory, the newest observation information is in full use. Thus the estimation errors of the new algorithm will be decreased noticeably, an accurate environment map will be established and high-precision localization will be obtained as well. The simulation results show that the location errors of x axis and y axis are both less than 1.5 m by the new algorithm. The estimating accuracy of the new algorithm is higher than that of SRCKF-SLAM, CKF-SLAM and EKF-SLAM algorithms. Adding different environmental noises, the position errors of ISRCKF are the smallest.