【摘要】 视觉惯性里程计(VIO)和伪卫星已经广泛应用于室内环境定位中，但在实际应用中，二者各自都有明显的缺陷。视觉里程计依赖于实际场景，在景深变化明显和光照不均匀的环境下会产生粗差，而且误差会不可避免地随着时间累积，但是在相邻帧间能保证相对高精度的位姿测量。由于受到室内多径的影响，伪卫星室内定位的精度和可靠性很难保证。为增加室内定位的可靠性和稳定性，基于抗差LM非线性优化理论，本文主要研究利用视觉惯性里程计的相邻帧间高精度位姿测量和伪卫星融合的室内高精度定位技术。该算法不仅可以抵抗粗差，而且可以减弱不同传感器间权重设置不合理带来的影响。最后使用在室内环境下搭建的高精度动态捕捉设备对组合定位方法进行实验验证。试验结果表明，该方法不依赖回环即可消除视觉惯性里程计的累积误差，有效提高室内定位精度及可靠性。利用改进的LM算法融合后场景1和场景2,相对于VIO单独定位精度分别提高了59.0%和77.5%。更多还原

【Abstract】 Visual inertial odometer(VIO) and pseudo-satellite have been widely used in positioning indoors, but in practical applications, both approaches have their own limitations. The visual odometer depends on the actual positioning environments. Gross errors occur in environments with obvious changes in depth of field and uneven illumination, and errors will inevitably accumulate over time. However, relatively high-precision pose measurements can be obtained between adjacent frames. Due to the influence of indoor multipath, the accuracy and reliability of pseudolite indoor positioning are difficult to guarantee. To increase the reliability and stability of indoor positioning, based on the robust LM nonlinear optimization theory, this study mainly investigate indoor high-precision positioning technology approach of integrating high-precision pose measurements of VIO between adjacent frame and pseudolite. The algorithm can not only resist gross errors, but also reduce the influence of unreasonable weight settings among different sensors. Finally, the high-precision dynamic capture equipment built in the indoor environment is used to verify the proposed method. The experimental results show that the method can eliminate the cumulative error of the visual inertial odometer without relying on the loopback, and effectively improve the indoor positioning accuracy and reliability. Compared with the VIO, the positioning accuracy is improved by 59.0% and 77.5% respectively after using the improved LM algorithm for scenes 1 and 2.更多还原