The assembly and manufacture of aerospace structures, in particular legacy products, relies in many cases on the skill, or rather the craftsmanship, of a human operator. Compounded by low volume rates, the implementation of a fully automated production facility may not be cost effective. A more efficient solution may be a mixture of both manual and automated operations but herein lies an issue of human error when stepping through the build from a manual operation to an automated one. Hence the requirement for an advanced automated assembly system to contain functionality for inline structural quality checking. Machine vision, used most extensively in manufacturing, is an obvious choice, but existing solutions tend to be application specific with a closed software development architecture.
Here we address these issues, presenting a robust solution for structural quality inspection using a low cost RGB-D sensor (Asus Xtion Pro Live) and open source libraries for point cloud analysis and processing. The system checks the quality of manually assembled sub-components before automated robot controlled operations are undertaken.
