This research developed an intelligent building exterior tile inspection robot, designed based on the movement patterns of geckos and inchworms. The front and rear tapping mechanisms are inspired by the pecking motion of woodpeckers. The robot is equipped with AI vision recognition and AI acoustic recognition, using a tiny AI camera and tapping device to detect defects. TinyML and CNN technologies are employed for data processing, and the system can transmit inspection results in real-time. This robot is the first in Taiwan to apply tiny machine learning to building exterior inspection.

The innovative features of this study＇s GLEWBOT include: bio-inspired design combining gecko adhesion and woodpecker tapping techniques; real-time audio analysis using TinyML; integration of CNN for defect classification; autonomous defect recognition; modular design and intelligent obstacle avoidance system for improved efficiency and safety. Its innovation has been recognized by platforms such as Circuit Digest, Arduino Blog, and Hackster.io.

This technology＇s industrial potential includes applications in buildings, bridges, offshore wind turbines, and oil pipelines. Its automation, precision, and efficiency can reduce maintenance costs, increase work efficiency, and decrease risks. Widespread use of this innovation will advance inspection industry technology.