A Simplified Magnetic Positioning Approach Based on Analytical Method and Data Fusion for Automated Guided Vehicles
A Simplified Magnetic Positioning Approach Based on Analytical Method and Data Fusion for Automated Guided Vehicles
复制标题
DOI：
10.1109/tmech.2021.3106679
复制
发表时间：
2022-10
期刊：
影响因子：
--
通讯作者：
Houde Dai;Pengfei Guo;Shijian Su;Shuang Song;Silin Zhao;Shuying Cheng
中科院分区：
文献类型：
--
作者：
Houde Dai;Pengfei Guo;Shijian Su;Shuang Song;Silin Zhao;Shuying Cheng
磁力定位方法是相对固定路径条件下自动导引车的一种可靠解决方案。铁氧体磁体通常埋在地下，用作AGV停车和转向的磁钉(MN)。我们以前的研究实现了一种基于超强NdFeB磁体的新型MPA，并通过磁偶极子模型最小化了磁强计数据与预测值之间的迭代，具有比传统MPA更优异的定位性能。然而，迭代优化算法依赖于初始猜测设置；此外，基于多个磁强计数据的复杂计算使其难以在微控制器上实时执行。在本文中，我们在分析方法的基础上提出了一种简化的多目标规划模型。从磁偶极子模型出发，推导出计算MN二维位置的解析表达式。即使是单一的三轴磁力仪也可以定位MN，而多个磁力仪往往可以实现更高的定位性能。因此，基于置信度准则对多台磁强计的定位结果进行融合。与基于Levenberg-MarQuardt算法的定位算法相比，该算法可以获得与之相当的定位精度，但具有更低的计算复杂度和更强的鲁棒性。因此，本文提出了一种用于AGV应用的高精度多点跟踪的新概念，而不需要复杂的视觉感知。
Magnetic positioning approach (MPA) is a reliable solution for automated guided vehicles (AGVs) under relatively fixed route conditions. Ferrite magnets are usually buried in the ground and employed as magnetic nails (MNs) for the parking and steering of AGVs. Our previous studies implemented a novel MPA based on super-strong NdFeB magnets and the minimize iterations between magnetometer data and predicted values via the magnetic dipole model, with the advantage of more exceptional positioning performance than traditional MPAs. Nevertheless, the iterative optimization algorithms depend on the initial guess setting; besides, the complicated calculation based on multiple magnetometer data makes it difficult to be executed on a microcontroller in real-time. In this article, we propose a simplified MPA based on the analytical method. The analytical expression for calculating the two-dimensional position of an MN is derived from the magnetic dipole model. Even a single tri-axis magnetometer can localize the MN, whereas multiple magnetometers tend to achieve higher positioning performance. Thus, positioning results of multiple magnetometers are fused based on the confidence criterion. Comparing with the MPA based on the Levenberg–Marquardt algorithm, the proposed MPA could achieve comparable positioning accuracy but with far lower computational complexity and greater robustness. Thereby, this article introduces a new concept of high-precision MPA for AGV applications without the complicated visual perception.