一种单目视觉里程计/UWB组合室内定位方法  

A monocular visual odometry/UWB combination indoor positioning method

作　　者：徐爱功[1] 闫可新 高嵩 宋帅 隋心[1] XU Aigong;YAN Kexin;GAO Song;SONG Shuai;SUI Xin(School of Geomatics,Liaoning Technical University,Fuxin,Liaoning 123000,China;South High-speed Railway Research and Development Center,Guangzhou South Surveying and Mapping Technology Co.Ltd.,Guangzhou 510000,China)

机构地区：[1]辽宁工程技术大学测绘与地理科学学院,辽宁阜新123000 [2]广州南方测绘科技股份有限公司南方高铁研发中心,广州510000

出　　处：《导航定位学报》2021年第5期82-88,共7页Journal of Navigation and Positioning

基　　金：辽宁省重点研发计划项目(2020JH2/10100044);辽宁省自然科学基金计划指导计划项目(2019-ZD-0051)。

摘　　要：针对室内环境下,单目视觉里程计定位中存在的尺度不可观和误差累积、超宽带(UWB)定位中存在的易受非视距误差影响的问题,提出一种单目视觉里程计/UWB组合定位方法,借助UWB获得绝对的位置信息,利用最小二乘法对单目视觉里程计进行绝对尺度估计;借助单目视觉里程计的位移增量,设计UWB非视距误差鉴别方法。然后将单目视觉里程计的位移增量和UWB的测距信息作为量测值,利用扩展卡尔曼滤波算法进行数据融合,实现室内定位。实验结果表明,所提算法能够有效地减小单目视觉里程计的误差累积和剔除UWB的非视距误差,提高组合定位系统的精度和稳定性。Aiming at the problems of unmeasurable scale and error accumulation in the positioning of monocular visual odometry in indoor environment, as well as the problem of being easily affected by Non-Line-of-Sight(NLOS) error in UltraWide Band(UWB) positioning, a monocular visual odometry/UWB combined positioning method was proposed. With the help of UWB, the absolute position information was obtained, and the least square method was used to estimate the absolute scale of monocular visual odometry. With the aid of the displacement increment of the monocular visual odometry, the threshold was set for NLOS error identification of UWB ranging value. Then, the displacement increment of monocular visual odometry and the ranging information of UWB were taken as measurement values, and the extended Kalman filtering algorithm was used for data fusion to realize indoor positioning. The experimental results show that the proposed algorithm can effectively reduce the error accumulation of monocular visual odometry and eliminate the NLOS error of UWB, and improve the accuracy and stability of the combined positioning system.

关 键 词：视觉里程计 超宽带 尺度估计 非视距误差 室内定位 

分 类 号：P228[天文地球—大地测量学与测量工程]