Scholarlink

<p>A methodology is presented for the digital reconstruction of an underground tunnel geometry using UAV photogrammetry, which is first tested in a computer simulation environment using the MRS UAV System, the Robot Operating System (ROS), and the Gazebo open-source robotics simulator. An algorithm for UAV navigation inside the tunnel is proposed, aiming to maintain a centralized position relative to the walls, ceiling, and floor. The proposed navigation algorithm can be used in a tunnel environment with few obstacles and overhanging structures, characterized by a practically constant cross-section along its entire length, which renders localization algorithms that utilize LiDAR scanning and point clouds impractical. These characteristics are common to rail or road transport tunnels. The methodology is tested in a computer simulation, and photogrammetry results showed that it is possible to digitally reconstruct the reference underground tunnel and faithfully reproduce details in texture, shape, and color. Thus, experiments were carried out in sections of a highway tunnel under construction to apply the same methodology for reconstructing three-dimensional geometry using photogrammetry, the images of which come from cameras onboard a drone with autonomous navigation using the same algorithm. The quantitative evaluation revealed a cross-sectional area difference of 0.21% between the designed area (70.72 <InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(m^2\)</EquationSource> </InlineEquation>) and the area obtained from the photogrammetric model (70.57 <InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(m^2\)</EquationSource> </InlineEquation>), confirming the high precision. Qualitatively, the model effectively represented textures and colors, validating the methodology for real-world applications.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Methodology for tunnel inspection using drone with autonomous navigation

  • 期刊论文

  • 发表日期：

    2026年2月11日

查看全文
  • Leandro Silva de Assis,
  • Antonio Carlos Daud Filho,
  • Lidia Gianne Souza da Rocha,
  • Kelen Cristiane Teixeira Vivaldini,
  • Glauco Augusto de Paula Caurin,
  • Marcos Massao Futai

摘要

A methodology is presented for the digital reconstruction of an underground tunnel geometry using UAV photogrammetry, which is first tested in a computer simulation environment using the MRS UAV System, the Robot Operating System (ROS), and the Gazebo open-source robotics simulator. An algorithm for UAV navigation inside the tunnel is proposed, aiming to maintain a centralized position relative to the walls, ceiling, and floor. The proposed navigation algorithm can be used in a tunnel environment with few obstacles and overhanging structures, characterized by a practically constant cross-section along its entire length, which renders localization algorithms that utilize LiDAR scanning and point clouds impractical. These characteristics are common to rail or road transport tunnels. The methodology is tested in a computer simulation, and photogrammetry results showed that it is possible to digitally reconstruct the reference underground tunnel and faithfully reproduce details in texture, shape, and color. Thus, experiments were carried out in sections of a highway tunnel under construction to apply the same methodology for reconstructing three-dimensional geometry using photogrammetry, the images of which come from cameras onboard a drone with autonomous navigation using the same algorithm. The quantitative evaluation revealed a cross-sectional area difference of 0.21% between the designed area (70.72 \(m^2\) ) and the area obtained from the photogrammetric model (70.57 \(m^2\) ), confirming the high precision. Qualitatively, the model effectively represented textures and colors, validating the methodology for real-world applications.