HyBridGen – AI-based Bridge Generator

Bridge design has so far been strongly shaped by individual engineering expertise, resulting in quality variations and lengthy planning processes. Additionally, early citizen participation can be challenging to achieve. The HyBridGen project therefore develops a largely automated bridge generator that creates plausible and feasible designs from existing route and geodata combined with engineering knowledge, supported by AI. These designs are photorealistically visualized and interactively explained, thereby enabling transparent decision-making.
A central foundation is an ontology that structures knowledge about bridges and their boundary conditions. Based on design criteria (e.g., environmental data, load capacity, budget), machine learning generates rapid, valid designs that are provided in a parameterized form. Validation is carried out in a pilot project in the Central German mining district (Saxony-Anhalt).