Magneto-Adaptive IoT-Enabled Smart Cleaning Robot with AI-based Dirt Detection

Glass façade cleaning in high-rise buildings poses
challenges due to safety risks, high labor costs, and
inconsistent manual results. This study presents a Magneto-
Adaptive IoT-Enabled Smart Glass Cleaning Robot as a
safer, more efficient alternative. The robot uses neodymium
magnets for stable adhesion to metal or metal-framed glass
and servo-actuated motors for precise movement. A
Raspberry Pi controller, integrated with an HD camera,
runs the YOLOv8 algorithm to detect heavily soiled areas.
Upon detection, a water spray and rotating brush
mechanism activates to perform targeted cleaning. IoT
capabilities allow real-time monitoring and control via a
mobile app, including manual override and emergency stop
functions. Experimental evaluation shows the model
achieved an Intersection over Union (IoU) of 64%, precision
of 83%, recall of 71%, and an F1-score of 76.5%, indicating
balanced detection performance. The proposed system
offers improved cleaning coverage, operational safety, and
energy efficiency for modern high-rise glass structures.