| 摘要: |
| 面向非视距(Non-line-of-sight，NLOS)场景中的人员定位需求，针对超宽带(Ultra-wideband，UWB)定位技术受多径信号干扰而精度下降的问题，提出了一种UWB误差补偿与视觉检测的融合定位系统。采用二维格网建模方法，设计实验获取场景各点UWB信号的到达时间误差。使用误差数据训练袋装树模型，预测并补偿多径信号对定位结果的干扰。结合工业相机拍摄的图像，提出新的视觉检测辅助定位方法。基于图像差分和形态学运算检测人员在图像中的位置，映射为实际坐标并采用松组合方式与UWB定位结果融合。实验结果表明，该融合定位系统相较于传统UWB定位技术，最大静态定位误差由0.54 m减小至0.21 m，最长定位丢失距离由4.4 m降低至0.36 m。 |
| 关键词: 室内定位 UWB定位 UWB误差补偿 视觉检测 |
| DOI：10.20079/j.issn.1001-893x.241226001 |
|
| 基金项目: |
|
| A Fusion Location System Based on UWB Error Compensation and Vision Detection |
| SI Linshun,LIU Mingqian,ZHANG Pengfei,LIU Zhiguo,XUE Renkui |
| (1.Shandong Nuclear Power Conpany Ltd.，Haiyang 265100，China;2.Beijing Orient Institute of Measurement and Test，Beijing 100080，China) |
| Abstract: |
| According to the human positioning requirement in non-line-of-sight(NLOS) scenarios，a positioning system based on the fusion of ultra-wideband(UWB) and image processing is developed to address the issue of positioning accuracy degradation in UWB positioning technology caused by multipath signal interference.The time-of-arrival error in the scenario is determined based on the 2D grid modeling method，and the bagged Classification And Regression Tree(CART) model is constructed to predict and compensate the positioning error caused by multipath interference.By using the images captured by camera，a novel vision-aided positioning method is proposed based on the image difference and morphological operation to detect personnel positions.Then the position coordinates are mapped to actual coordinates and fused with UWB positioning result through a loosely-coupled integration framework.The experimental results illustrate that compared with that of the traditional UWB system，the maximum static positioning error of the proposed system is reduce from 0.54 m to 0.21 m，and the longest signal loss distance is reduced from 4.4 m to 0.36 m. |
| Key words: indoor positioning UWB location UWB error compensation visual inspection |
《电讯技术》编辑部 