A Wireless Local Positioning System Concept and 6D Localization Approach for Cooperative Robot Swarms Based on Distance and Angle Measurements
A Wireless Local Positioning System Concept and 6D Localization Approach for Cooperative Robot Swarms Based on Distance and Angle Measurements
复制标题
DOI：
10.1109/access.2020.3004651
复制
发表时间：
2020
期刊：
影响因子：
3.9
通讯作者：
Johanna Geiss;Erik Sippel;Patrick Gröschel;M. Hehn;Martin Schütz;M. Vossiek
中科院分区：
文献类型：
--
作者：
Johanna Geiss;Erik Sippel;Patrick Gröschel;M. Hehn;Martin Schütz;M. Vossiek
在无线传感器网络中，空间分布的节点提供位置相关的传感器信息。因此，关于所有节点的3D位置的知识对于需要自主移动性的众多应用至关重要。此外，为了获得节点的姿态和完整的6D网络星座，还需要每个节点的3D定向。虽然无线传感器网络存在许多理论上的定位概念，但仍然缺乏可靠的系统和定位概念，从而能够在现实世界的场景中进行鲁棒的实时跟踪。因此，我们提出了一个系统的方法，基于先进的24 GHz无线本地定位系统，提供对节点之间的距离和角度测量。此外，提出了一种基于扩展卡尔曼滤波的定位算法，该算法评估这些测量值以跟踪网络中所有节点的时变6D姿态。因为只有相对测量是可用的，所以选择一个节点来定义联合导航系统。因此，所提出的系统在没有任何先前安装的基础设施或网络的先前信息的情况下工作。系统和定位算法进行了验证，在一个移动的无线传感器网络中进行的测量，包括六个节点在室内的情况下，具有较强的多径传播。然而，尽管环境充满挑战，该系统仍然可以对网络中的所有机器人进行稳定和准确的6D姿态估计，3D定位均方根误差为6至15厘米。
In wireless sensor networks, spatially distributed nodes provide location-dependent sensor information. Therefore, knowledge about the 3D position of all nodes is crucial for the numerous applications that require autonomous mobility. Furthermore, to acquire the nodes’ poses and the complete 6D network constellation, the 3D orientation of each node is also required. While many theoretical localization concepts exist for wireless sensor networks, there is still a lack of reliable system and localization concepts which enable robust real-time tracking in real-world scenarios. Therefore, we present a system approach based on an advanced 24 GHz wireless local positioning system, providing distance and angle measurements between pairs of nodes. Furthermore, an extended Kalman filter based localization algorithm is proposed, which evaluates these measurements to track the time varying 6D poses of all nodes in the network. Because only relative measurements are available, one node is chosen to define a joint navigation system. Hence, the proposed system works without any previously installed infrastructure or prior information of the network. The system and localization algorithm are validated by measurements performed in a mobile wireless sensor network comprising six nodes in an indoor scenario with strong multipath propagation. However, despite the challenging environment, the system allows for a stable and accurate 6D pose estimation of all robots in the network with 3D positioning root mean square errors of 6 to 15cm.