Automated quality inspection for the manufacturing of highly customized building modules

The manufacturing of prefabricated building modules is a highly customized process. Each module consists of a base steel frame with numerous support struts and other attachments added. Therefore, an elaborated inspection process is required to ensure the presence and correct placement of all parts. This process is currently performed manually, which makes it very time consuming and prone to human error, such as overlooking incorrectly placed or missing components. To address these challenges, we propose an innovative solution for a highly automated inspection system. Our approach uses point clouds, generated by a 3D laser scanner placed around the building module, to compute a match with a CAD model of the module and its parts. Instead of a direct scan-CAD comparison, we utilize a novel method of comparing the point clouds to a simulation of the expected sensor data, generated from their known locations in relation to the module. This results in an enhanced accuracy of the inspection in case of misplaced or swapped parts. Furthermore, the generation of an inspection report is automated as well. This report includes views of every relevant part, which are generated automatically by simulating a virtual camera to determine its optimal placement for an informative image. The report is presented to the auditor via a web interface, allowing for an intuitive assessment of the inspection results and significantly reducing the potential for overlooked manufacturing faults.