Relay-based Railways Interlocking Systems (RRIS) carry out critical functions to control stations. Despite being based on old and hard-to-maintain electro-mechanical technology, RRIS are still pervasive. A powerful CAD modeling and analysis approach based on symbolic logic has been recently proposed to support the re-engineering of relay diagrams into more maintainable computer-based technologies. However, the legacy engineering drawings that need to be digitized consist of large, hand-drawn diagrams dating back several decades. Manually transforming such diagrams into the format of the CAD tool is labor-intensive and error-prone, effectively a bottleneck in the reverse-engineering process. In this paper, we tackle the problem of automatic digitalization of RRIS schematics into the corresponding CAD format with an integrative Artificial Intelligence approach. Deep learning-based methods, segment detection, and clustering techniques for the automated digitalization of engineering schematics are used to detect and classify the single elements of the diagram. These elementary elements can then be aggregated into more complex objects leveraging the domain ontology. First results of the method’s capability of automatically reconstructing the engineering schematics are presented.
