城市环境下伪卫星/地标/IEZ图优化定位算法

The graph optimization positioning algorithm of pseudolite/landmarks/IEZ for urban environment

关键词:  位姿图优化  多场景识别  室内外定位  城市环境  伪卫星定位

Key words:  Pose graph optimization  Multi-scene recognition  Indoor and outdoor positioning  Urban environment  Pseudolite positioning

摘要:

针对城市环境行人室内外定位精度和连续性差的问题，提出了一种伪卫星/地标/IEZ融合的行人位姿图优化定位方法，采用位姿图对伪卫星/地标/IEZ的融合定位结果进行优化处理。构建了室内外多场景识别机制，以判断行人当前所处场景，辅助实现融合方式的自主切换。通过引入差异进化优化搜索框架，提出了新型伪卫星室内大范围高精度定位算法。实验结果表明，伪卫星/地标/IEZ位姿图优化融合定位方法较传统扩展卡尔曼滤波（EKF）融合定位方法具有更高的定位精度和连续性；与遍历搜索算法和多普勒差分算法相比，新型伪卫星室内定位误差分别减少了86.9%和89.95%；室内外多场景识别机制能够辅助实现不同融合定位方式的自主切换。

Abstract:

To address the problem of poor accuracy and continuity of pedestrian positioning in urban indoor and outdoor environments, a pedestrian pose graph optimization positioning method based on the combination of pseudolite/landmark/IEZ fusion is proposed. The pose graph is used to optimize the fusion positioning results of pseudolite/landmark/IEZ. An indoor and outdoor multi-scene recognition mechanism is built to determine the current scene of pedestrians and support the autonomous switching of fusion modes. A novel pseudolite indoor wide-range high-precision positioning algorithm is proposed by introducing a differential evolution optimization search framework. The experimental results show that the pseudolite/landmark/IEZ pose graph optimization fusion positioning method has higher positioning accuracy and continuity than the traditional EKF fusion method. Compared with the traversal search algorithm and the Doppler difference algorithm, the positioning errors of the novel pseudolite indoor positioning algorithm were reduced by 86.9% and 89.95% respectively. The indoor and outdoor multi-scene recognition mechanism can help to achiev autonomous switching of different fusion positioning methods.