Automated infrastructureless inspection of construction sites -

Abstract

Site inspection is a notably tedious, time-consuming, and error-prone process when carried out manually by construction inspectors. Hence, it is a prime candidate for automation, which would reduce the effort and time incurred while improving the accuracy and organization of the data obtained. Incorporating recent contributions to the field of laser-based measurement systems, the following thesis presents a system aimed at automating the inspection process of a construction-site: the comparison of as-built vertical surfaces to their as-designed counterparts. The system relies on a Rao-Blackwellized Particle Filter occupancy grid SLAM to estimate the motion of the inspector and build a dense map of the environment, which is then used to automatically validate the original 2D architectural designs. In addition the proposed system does not depend on a pre-installed or dedicated sensor infrastructure and is thus independent from the infrastructure of the construction site. The components of the developed system have been tested through proof-of-concept experiments and results highlight the potential of using laser measurement data as proposed, for improving the inspection processes of wall positioning in construction engineering applications.