一种单目VIO/UWB室内组合定位方法

A monocular VIO/UWB indoor integrated positioning method

针对单目视觉惯性里程计(VIO)在室内环境下易漂移、定位误差较大的问题,提出了一种基于单目VIO/超宽带(UWB)组合的室内高精度定位方法。该方法结合单目VIO输出的位置信息和UWB的测距信息,利用自适应卡尔曼滤波进行数据融合,考虑到UWB测距值易受非视距(NLOS)误差的影响,通过新息向量构建组合系统的抗差模型,以减小异常测距值对组合系统的影响,同时引入萨格-胡萨(Sage-Husa)滤波对系统噪声进行实时估计和修正。实验结果表明该定位方法能够有效缓解单目VIO的累积漂移和UWB的非视距误差,相较于基于扩展卡尔曼滤波器(EKF)的单目VIO/UWB组合方法,其平面均方根误差降低了52.3%,有效提高了组合系统的定位精度和鲁棒性。

Aiming at the problems of monocular visual inertial odometry(VIO) easy to drift and large positioning error in indoor environment, this paper proposes an indoor high-precision positioning method based on monocular VIO/ultra-wide band(UWB) integration. This method integrates position information output by VIO and the ranging information about(UWB), and uses adaptive kalian filter for data fusion. Considering that the UWB ranging value is susceptible to non line of sight(NLOS)error, the robust model of the integrated system is constructed by the new interest vector to reduce the influence of the abnormal ranging value on the integrated system, and Sage-Husa filter is introduced to estimate and correct the system noise in real time.The experimental results show that the positioning method can effectively alleviate the cumulative drift of monocular VIO and the nonline-of-sight error of UWB. Compared with the monocular VIO/UWB integrated method based on extend Kalman filter(EKF), the root mean square error is reduced by 52.3%, and the positioning accuracy and robustness of the integrated system are effectively improved.