一种单目视觉ORB-SLAM/INS组合导航方法  

Integrated navigation method of monocular ORB-SLAM/INS

作　　者：周绍磊[1] 吴修振[1] 刘刚[1] 张嵘[2] 徐海刚 

机构地区：[1]海军航空工程学院控制工程系,烟台264001 [2]清华大学精密仪器系,北京100084 [3]北京自动化控制设备研究所,北京100074

出　　处：《中国惯性技术学报》2016年第5期633-637,共5页Journal of Chinese Inertial Technology

基　　金：国家自然科学基金(61004002)

摘　　要：针对惯性/卫星组合导航系统在卫星导航失效时无法使用的问题,提出了单目视觉ORB-SLAM/INS组合导航方法,用于扩展组合导航系统在强干扰环境和室内环境的应用范围。该算法分为两个阶段:初始化阶段,当ORB-SLAM形成闭环时设计算法在线估计单目视觉ORB-SLAM算法的尺度因子;导航阶段,ORB-SLAM系统输出的位置信息经过尺度变换后作为观测量进行卡尔曼滤波,估计INS导航系统的误差状态量从而修正惯导系统的误差。设计了硬件和软件平台对提出的组合导航方法进行试验验证。跑车实验结果表明:所设计的ORB-SLAM/INS组合导航系统具有较高的定位精度,导航时间6 min定位误差为1.162 m,且不随时间漂移,具有很强的应用价值。In order to solve such problem that an integrated INS/GNSS navigation system stops working in satellite failure situation, a deeply-integrated navigation method of monocular ORB-SLAM/INS is proposed to extend the application range of the integrated navigation system in severe interference environment and indoor circumstances. The method can be divided into two phases： in initialization phase, an algorithm is designed to estimate the scale factor online when the ORB-SLAM forms a closed loop; in navigation phase, the error of INS is corrected after the error states of INS are estimated through a Kalman filter whose observations is ORB-SLAM positions. The hardware and software platforms are designed to test the proposed integrated navigation method. Field test results indicate that the position accuracy of the integrated navigation system is satisfactory, with 1.162 m mean error in 6 min and no drift with time, showing that the proposed method has a notable application value.

关 键 词：ORB-SLAM INS 尺度因子 卡尔曼滤波 组合导航 

分 类 号：U666.1[交通运输工程—船舶及航道工程]