Development of a robot vision system for construction automation

Abstract

In construction projects, it is a crucial industry-wide practice to perform routine inspections on works to ensure that the buildings are delivered according to the projected timeline as well as the necessary standards of quality to be met. Inspection processes are laborious, time-consuming, and prone to human error or inconsistency. With recent developments in technology such as Artificial Intelligence and Building Information Models (BIM), and construction automation solutions can be implemented and integrated with data from BIM. In this paper, an Aerial Robot Vision System for inspection of indoor building installations such as architectural and electrical fixtures is implemented using a DJI Tello Aircraft to maneuver within a construction building for the collection of visual data. A YOLOv5 Detector performs detections of the component installations, which are then used for analysis based on infrastructure data extracted from BIM.