Automated Infrastructure Inspection Based on Digital Twins and Machine Learning

Abstract

One of the key challenges in our modern society is the provision of safe transport infrastructure. Infrastructure managers are subject to regulations requiring major infrastructures to be periodically checked for damage before it becomes a safety hazard. In the standard structural inspection, specially qualified civil engineers travel to the object to be inspected on site. Notes, sketches and photos are prepared for the subsequent report. The inspectors are introduced to non-directly accessible locations with special, heavy inspection equipment. During such a test the object is not or only partially usable, which leads to interruptions, delays, traffic jam and thus considerable non-availability costs. In recent years, approaches of a drone-based structural inspection are increasingly noticeable. These are mainly limited to a visual inspection of the created optical images. By using new technologies, a more objective and faster structural inspection can be carried out at a lower cost. In order to achieve the highest possible level of automation, the actual test is no longer performed on the real object, but on a digital twin of the construction. The assessment of damages and reporting is carried out automatically.