Developing a Quadruped Robot-Based 3D and Panoramic Thermal Mapping System: Moving towards Smart and Automated Building Thermal Inspection Using LiDAR-Based Digital Twins and Sensor Fusion Technologies

Robotic 3D thermal mapping (R3DTM) is of great importance for thermal inspection of as-built facilities to ensure building thermal performance for maintaining satisfaction of indoor comforts. Existing R3DTM systems commonly depend on unmanned ground vehicles and a single thermal camera. However, such systems are often constrained by the complex indoor layouts, such as stairs, and the limited field of view of the single thermal camera, resulting in low maneuverability of the entire system and inefficiency of the 3D thermal mapping process. As such, this paper proposes a novel R3DTM system integrating an image-stitching-based panoramic thermal camera system and a quadruped robot to improve the system’s maneuverability and thermal mapping efficiency. Experimental results showed that the developed system achieved smooth real-time images stitching at 9 frames per second and promising sensor calibration performance to fuse thermal image and LiDAR-based point clouds in support of thermal map generation. This study contributes to the body of knowledge by integrating panoramic thermography, LiDAR-based 3D capturing, and quadruped robotics to advance R3DTM towards digital twins-enabled building thermal inspection.