基于迭代平方根CKF的SLAM算法

摘要

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.%在大尺度环境中，平方根容积卡尔曼同步定位与地图构建算法的非线性误差严重制约了算法的定位精度，为解决这一问题，提出了一种基于迭代平方根容积卡尔曼滤波的改进算法，该算法结合迭代理论，对平方根容积卡尔曼滤波的量测更新过程进行迭代更新，充分利用最新的观测信息，降低滤波的估计误差，从而构建精确的地图并获得高精度的定位信息。仿真实验结果表明，采用本算法后， x轴和y轴方向上的位置误差均在1�5 m以内，估计结果明显优于SRCKF－SLAM、CKF－SLAM和EKF－SLAM算法；添加不同的环境噪声后进行仿真实验，该算法所取得的位置误差相比仍是最小的。利用该算法可以有效地减小非线性误差造成的影响，提高SLAM的定位精度。