P2P-DroneLoc: Peer-to-Peer Localization for GPS-Denied Drones using Camera and WiFi Fine Time Measurement
P2P-DroneLoc: Peer-to-Peer Localization for GPS-Denied Drones using Camera and WiFi Fine Time Measurement
复制标题
DOI：
10.1109/ants56424.2022.10227756
复制
发表时间：
2022-12
期刊：
影响因子：
--
通讯作者：
Khadija Ashraf;A. Ashok
中科院分区：
文献类型：
--
作者：
Khadija Ashraf;A. Ashok
在这项工作中，我们设计，实施和评估无人机辅助协调定位系统，以帮助导航远程操作的全球定位系统（GPS）拒绝无人机。在没有任何基站或中央实体的要求的情况下，在该系统中，启用GPS的辅助无人机向低功能遇险无人机提供定位帮助。当发现遇险无人机在附近飞行时，助手无人机估计遇险无人机相对于其自身位置的绝对GPS位置信息。这是通过计算遇险无人机与辅助无人机的相对位置来执行的。然后将估计的位置传送给遇险无人机。在本文中，我们评估了两种基本方法的范围估计无人机之间的对等相对定位：（a）相机和计算机视觉投影理论，（B）WiFi精细时间测量（FTM）。我们设想，助手无人机配备摄像头和WiFi FTM接入点（AP），而遇险无人机仅配备WiFi FTM AP。我们通过三个估计指标的准确性来评估我们提出的点对点定位：（i）无人机之间的范围或距离，（ii）GPS方位（角度），以及（iii）GPS位置（坐标）。基于我们对由户外静态和飞行无人机设置组成的数据集的分析，我们的方法可以实现1- 4米范围内的中值定位精度。
In this work we design, implement, and evaluate a drone assisted coordinated localization system to help navigate remotely operating Global Positioning System (GPS) denied drones. Without any requirements of a base station or a central entity, in this system, a GPS enabled helper drone offers localization help to low functionality distress drones. The helper drone when discovers a distressed drone flying in proximity, estimates the distress drone’s absolute GPS location information respective to its own location. This is performed by computing the relative position of the distress drone with the helper drone. The estimated location is then communicated to the distress drone. In this paper, we evaluate two fundamental approaches for range estimation for peer-to-peer relative positioning between drones: (a) camera and computer vision projection theory, and (b) WiFi Fine Time Measurements (FTM). We envision that a helper drone equips a camera and a WiFi FTM access point (AP), while a distress drone equips only a WiFi FTM AP. We evaluate our proposed peer-to-peer localization via accuracy across three estimated measures: (i) range or distance between the drones, (ii) GPS bearing (angle), and (iii) GPS location (coordinates). Based on our analysis of our dataset consisting of outdoor static and flying drones setup, our methods result in a median localization accuracy within 1-4m.