摘要
为满足高性能自主导航的要求,解决单一惯性导航系统定位误差不断积累的问题,考虑利用相机元件采集的视觉图像信息校正定位解。提出一种基于无损卡尔曼滤波的多速率数据融合算法,在有图像信息的每个时刻利用多视图的几何约束来校正,有效解决单一惯性导航定位的漂移问题;分析多速率数据融合的算法,并与普通多传感器同步融合的单速率算法做出比较;最后采用车载传感器采集的数据进行实验。实验结果表明视觉辅助惯性导航能够有效地抑制单一惯性导航定位漂移的问题,并且与单速率数据融合算法相比,该多速率数据融合算法的定位精度更高。
In order to meet the requirements of high performance of the autonomous navigation and solve the problem of continuous accumulation of positioning error of single inertial navigation system, the paper used the visual information collected by camera components to correct the positioning solution, and proposed an algorithm of multirate data fusion based on UKF: the geometric constraints of multi views were used to do the correction at each moment of the visual information, which could effectively solve the drifting problem of single inertial navigation system; and the algorithm of multirate data fusion was analyzed; then, a comparison between the single rate data fusion of common multi sensor synchronization and multirate data fusion was made; finally, experimental data were collected by sensors mounted in a vehicle. Results showed that the inertial with visual aided navigation could effectively suppress the problem of positioning drift with single inertial navigation, and the multirate data fusion would be more precise than the single rate data fusion in positioning.
作者
王晨琳
刘海颖
蒋鑫
WANG Chenlin;LIU Haiying;JIANG Xin(College of Astronautics,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China)
出处
《导航定位学报》
CSCD
2018年第4期19-23,28,共6页
Journal of Navigation and Positioning
基金
江苏省自然科学基金项目(BK20161490)
关键词
视觉信息
惯性导航
无损卡尔曼滤波
多速率数据融合
visual information
inertial navigation
unscented Kalman filter
multirate data fusion