Robot-based inspection of civil infrastructure using frame semantics and linguistic metamodels

Motivation and research problem

Future construction sites, frequently referred to as "smart construction sites", will be characterized by digitalization, automation, and interconnection. Subsumed under "Construction 4.0", smart construction sites realize the fourth revolution in construction, taking advantage of the momentum gained from the progress in robotic systems research and development. Robotic systems are increasingly deployed to automate inspections of civil infrastructure both in the construction phase and in the operation phase. To conduct robot-based inspections, inspection routes must be calculated and, if necessary, the routes of robotic fleets, collaboratively conducting inspection tasks, must be minimized and efficiently distributed. Since using digital building models, i.e. building information models, is mandatory in many areas of public construction, it is reasonable to use building information models as a conceptual basis for calculating inspection routes. Currently, however, it is hardly possible to fully integrate robot information into digital building models, which would enable efficient communication and coordination among robotic systems.

Research objectives and expected results

This research project aims to provide a methodology for automating robot-based inspections using digital building models as a conceptual basis. The research hypothesis, formulated to achieve this aim, can be reduced to the following simple statement: Once an inspection task is specified, the inspection routes for robot fleets can reliably be calculated, based on (i) robot information and (ii) building information derived from digital building models. In this project, the semantics inherent to digital building models, which implicitly contain the information relevant to route calculation, serve as a formal basis for coupling robot and building information. Taking into account existing knowledge sources, the coupling is performed through frame semantics as a formal means for knowledge representation, since frame semantics are linguistically founded in cognitive science and have proven to be a powerful tool in construction grammar and computational linguistics. The frame-based knowledge, as a cognitive representation of the real world, is then mapped into linguistic metamodels in the form of ontologies, easily comprehensible by engineers, which allow generating specific inspection models. Using a reference bridge, standard-compliant building information models will be applied to validate the methodology. As a result, the research project is expected to provide a formal approach for reliably and efficiently deploying robotic systems to inspect civil infrastructure. It is further expected to provide a robust scientific foundation for assessing the potential of robotic systems to enhance the current state of the art of civil infrastructure inspections.


Professor Dr. Kay Smarsly
Hamburg University of Technology
Institute of Digital and Autonomous Construction
Blohmstraße 15
21079 Hamburg