Improved Vehicle Localization Based on Moving Horizon Estimation with Node Constraints
Improved Vehicle Localization Based on Moving Horizon Estimation with Node Constraints
复制标题
基于带节点约束的移动地平线估计的改进车辆定位
DOI：
10.1109/sii52469.2022.9708878
复制
发表时间：
2022
期刊：
影响因子：
--
通讯作者：
K. Sekiguchi
中科院分区：
文献类型：
--
作者：
Ryusei Toma;K. Nonaka;K. Sekiguchi
在本研究中，利用嵌入在移动地平线估计（MHE）中的节点约束，提高了全球导航卫星系统（GNSS）车辆定位的精度。在卫星信号未到达或出现多径的区域，GNSS测量结果会产生偏差，导致定位误差较大。在本研究中，通过MHE融合GNSS、里程计和地图信息来解决这一问题。MHE结合约束来提高估计精度。在本研究中，利用地图上的节点作为约束，在不需要高精度地图的情况下，对每个点使用节点修改估计位置。此外，MHE还可以在视界内修正过去的估计，修正一系列的估计位置，使其与运行轨迹一致。此外，通过利用节点上的位移估计偏差本身来修正GNSS偏差引起的估计误差。即使在节点处的约束超出视界后，这种修正仍然有效。我们进行了车辆定位的模拟，其中节点被放置在轨迹上。验证了该方法对GNSS定位误差的抑制作用。因此，以节点为约束的MHE可以改善受GNSS定位误差影响的估计。
In this research, the accuracy of vehicle localization using Global Navigation Satellite System (GNSS) is improved by utilizing node constraints embedded in Moving Horizon Estimation (MHE). In the area where the satellite signal does not reach, or multipath appears, GNSS measurements will be biased and result in large localization error. In this study, GNSS, odometry, and map information are fused by MHE to address this issue. MHE incorporate constraints to improve estimation accuracy. In this study, nodes on the map are utilized as constraints to modify the estimated positions using nodes for each point without requiring a high-precision map. In addition, MHE can modify the past estimation within the horizon, which correct the series of estimated position so that they coincide with the running trajectory. Furthermore, the estimation error due to GNSS bias is corrected by estimating the bias itself using the displacement at the node. This correction is effective even after the constraint at nodes goes out of the horizon. We conducted simulations of vehicle localization in which nodes were placed on the trajectory. It was verified that the proposed method suppressed the GNSS positioning error. Therefore, MHE with nodes as constraint could improve the estimation affected by the positioning error of GNSS.