迭代中心差分粒子滤波的SLAM算法

A SLAM algorithm based on an iterated central difference particle filter

为了提高SLAM算法中的位姿估计精度,通过在广泛使用的RBPF粒子滤波器中,利用迭代中心差分卡尔曼滤波器(ICDKF)来代替其中的扩展卡尔曼滤波器(EKF),并融合新的观测数据使提议分布更加接近后验概率分布,并且能够精确估计智能车辆的位姿,进而采用ICDKF算法更新特征地图的位置.该算法在保证定位精度的同时减少了计算的复杂度,提高系统的估计性能,增加了迭代算法的稳定性.仿真实验结果验证了迭代中心差分粒子滤波SLAM算法的有效性.

In order to improve the accuracy of position estimation in the simultaneous localization and mapping（SLAM） algorithm,this paper provided an iterated central difference Kalman filter（ICDKF） to compute the proposed distribution in the widely used Rao-blackwellized particle filter（RBPF） instead of the extended Kalman filter（EKF）.Furthermore,by combining new observation data,the proposed distribution was moved closer to the posteriori probability,and the position of the intelligent vehicle was accurately estimated,therefore updating the position of the feature map by the ICDKF.This algorithm decreased computational complexity,and improved the estimation performance of the system along with the stability of the iterated algorithm without decreasing accuracy.Simulation results validate the effectiveness of the proposed algorithm.