| 引用本文： |
-
徐正丽,肖素芳,杨明浩.基于无需校准激光雷达曲线与RGB-D图像的数据融合方法[J].广西科学,2024,31(5):1038-1048. [点击复制]
- XU Zhengli,XIAO Sufang,YANG Minghao.A Fusion Method Based on Uncalibrated LiDAR Curves and RGB-D Images[J].Guangxi Sciences,2024,31(5):1038-1048. [点击复制]
|
|
| 摘要: |
| 激光雷达和RGB-D摄像机是在各种机器人导航任务中广泛使用的两种传感器。尽管通过图像校准能够降低传感器中存在的噪声，但由于外部复杂环境的影响，融合数据中仍会有空洞和毛刺。为此，本文提出一种无需校准激光雷达曲线与RGB-D图像的数据融合方法。该方法利用激光雷达曲线和RGB-D图像数据在格式上存在显著差异但在深度信息上紧密相对的特点，通过时间配准和相关性分析，将激光雷达曲线在高度上与RGB-D图像配准。同时，即使在不进行校准的情况下，该方法也能将RGB-D图像在高度上的相应线条自动配准到激光雷达曲线在宽度上的范围。使用室内开放式机器人导航数据集Robot@Home中的数据对本文方法进行评估，结果表明，该方法可同时为激光雷达曲线和RGB-D图像的原始数据去噪。该方法已在真实的导航环境中得到了验证，可扩展应用于机器人导航的更精确的二维地图构建中。 |
| 关键词: 激光雷达 深度图像 数据融合 二维地图重建 机器人导航 |
| DOI：10.13656/j.cnki.gxkx.20241127.019 |
| 投稿时间：2023-11-08修订日期：2024-01-12 |
| 基金项目:国家自然科学基金项目(71463010，22180155466)，广西科技计划项目(2021GXNSFBA220048，桂科AB21220038)和桂林科技计划项目(20220115-1，20230110-1)资助。 |
|
| A Fusion Method Based on Uncalibrated LiDAR Curves and RGB-D Images |
|
XU Zhengli1, XIAO Sufang1, YANG Minghao2
|
| (1.Guilin University of Electronic Technology, Guilin, Guangxi, 541004, China;2.Institute of Automation of the Chinese Academy of Sciences, Beijing, 100190, China) |
| Abstract: |
| LiDAR and RGB-D cameras are two widely used sensors in various tasks of robot navigation.Although image calibration can reduce the noise present in both sensors,there may still be holes and burrs due to the complex external environment.To address these issues,a fusion method based on uncalibrated LiDAR curves and RGB-D images is proposed.The proposed method utilizes the significant differences in format but close alignment of depth information between LiDAR curve and RGB-D image data.Through time alignment and correlation analysis,two-dimensional LiDAR curves can be matched with RGB-D images at heights.Meanwhile,the proposed method can automatically match the corresponding lines in RGB-D images to the width range of the LiDAR curve even if calibration is not necessary.The proposed method is evaluated on the public indoor robot navigation database Robot@Home.The experimental results show that the proposed method simultaneously de-noises the raw data of LiDAR curves and RGB-D images.Moreover,the proposed method is validated in real navigation environments and can be applied to the reconstruction of more accurate 2D maps for robot navigation. |
| Key words: LiDAR depth image data fusion 2D map reconstruction robot navigation |
