基于HurberM极大似然估计求解的CKF滤波定位算法  

CKF Filtering Algorithm Based on HurberM Maximum Likelihood Estimation

作　　者：戴卿[1,2] DAI Qing(Department of Architectural Engineering,Chongqing Water Resources and Electric Engineering College,Chongqing 402102,China;Institute of Geospatial Information,Information Engineering University,Zhengzhou 450001,China)

机构地区：[1]重庆水利电力职业技术学院建筑工程系 [2]信息工程大学地理空间信息学院

出　　处：《控制工程》2018年第9期1760-1764,共5页Control Engineering of China

基　　金：渝水职院科研项目(K201708);重庆市教委科学技术研究项目(KJ1735452)

摘　　要：针对传统移动机器人定位算法精度欠佳的问题,设计了一种基于无线传感器网络滤波算法的移动机器人定位算法（HM_CKF）。该算法利用HurberM极大似然估计代价函数,将线性化后CKF观测矩阵求出,从而解决CKF滤波算法在未知非高斯白噪声干扰下估计精度欠佳的问题。然后在体育馆中基于WSNs网络构建了移动机器人定位实验环境,融合移动机器人动力学模型,对比了HM_CKF、CKF算法的定位精度。实验结果表明,在不含噪声干扰和含未知噪声干扰的两种情况下,HM_CKF算法定位精度依次比CKF算法提高了7%和15%。Aiming at the problem of poor accuracy of the traditional mobile robot localization algorithm, a mobile robot localization algorithm based on wireless sensor network filtering algorithm（HM_CKF） is designed. In this algorithm, the Hurb M maximum likelihood estimation is used to solve the problem that the CKF filter algorithm can be used to estimate the accuracy of the CKF filtering algorithm in the unknown non Gauss white noise. Then, based on the WSNs network, the mobile robot localization experiment environment is constructed, and the dynamic model of the mobile robot is integrated, and the positioning accuracy of CKF and HM_CKF algorithm is compared. The experimental results show that the accuracy of HM_CKF algorithm is improved by 7 % and 15 % compared with the CKF algorithm in the two cases without noise and with unknown noise.

关 键 词：无线传感器网络 HurberM代价函数 CKF滤波 移动机器人 定位 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]