Towards shared autonomy in construction of high-rise buildings: Mid-air façade panel detection and localization

This thesis explores how robots can help make construction sites safer and more efficient, especially when installing glass panels on tall buildings. It introduces a new computer vision system that allows a robot to “see” and understand where and how to grab glass panels in mid-air — even when visibility is poor due to wind, dust, or rain. Unlike existing methods, it doesn’t rely on machine learning or pre-training. The system works in real time, allowing robots and cranes to respond quickly and accurately. This research brings construction closer to safe, shared work between humans and intelligent machines.