Seamless Navigation, 3D Reconstruction, Thermographic and Semantic Mapping for Building Inspection

Area of Science:

• Geomatics Engineering
• Robotics
• Computer Vision

Background:

• Accurate 3D mapping and thermal analysis are crucial for infrastructure inspection and time-critical applications.
• Integrating indoor and outdoor spatial data presents significant technical challenges.
• Existing methods often lack real-time capabilities or seamless global referencing.

Purpose of the Study:

• To develop an automated workflow for real-time 3D thermal mapping and navigation in combined indoor/outdoor environments.
• To enhance inspection tasks and time-critical applications through integrated sensor data.
• To achieve seamless global referencing for georeferenced 3D point clouds.

Main Methods:

• Utilized a hand-held integrated positioning system (IPS) with visual-aided inertial navigation.
• Augmented the IPS with a passive thermal infrared camera and global referencing via surveyed optical markers (AprilTags).
• Fused stereo camera and thermal image data, calibrated a trifocal camera system, and integrated UAV data for a complete building hull point cloud.

Main Results:

• Generated a georeferenced 3D point cloud enriched with thermal intensity values.
• Achieved seamless fusion of terrestrial and aerial datasets for comprehensive building data.
• Demonstrated automated semantic 3D model reconstruction from fused sensor data.

Conclusions:

• The presented workflow enables automated, real-time 3D thermal mapping and navigation in complex environments.
• This approach significantly advances automated building information modeling (BIM) and inspection capabilities.
• The successful demonstration on a test site validates the workflow's practical applicability.