TwinGen

Participation in the TwinGen Research Project

The project, led by TUM, is dedicated to the development of approaches for the automated creation of digital twins for structural infrastructures. It focuses on methods for capturing existing structures using various methods and their transfer into digital models as a basis for operation and maintenance.

The project focuses on the development of processes to automatically generate digital twins of existing infrastructure structures (e.g. roads, bridges, hydraulic structures) as the basis for operation and maintenance. The sub-processes in development include the recording and processing of point clouds (e.g. using laser scanning or photogrammetry), extraction of geometric and semantic information using digital image processing, the integration of knowledge databases for semantic model generation and the evaluation of technical drawings using machine learning. The main activities are the development of algorithms, data structures and technical processes.

The aim is to develop processes with which high-quality digital twins can be generated for existing buildings in a largely automated manner. A major challenge is to extract the semantic information from the inventory documents and the inventory and to enrich the digital twin accordingly. Furthermore, methods are to be used to make the condition of a building assessable on the basis of images and point clouds.

The chair for computer-aided modeling and simulation focuses on the development of suitable parameterizations for buildings, optimization methods for model-to-cloud fitting and methods for extracting information from technical drawings in particular.
Furthermore aerial drones captured photogrammetric images of several bridges, in order to create point clouds of the photographed structures.

ZPP supports the scientific partners by introducing requirements from construction practice and their implementation. Furthermore, data bases are processed with the goal of cross-domain matching and corresponding concepts are introduced. Additionally, the scientific partners co-develop the infrastructure for semantic digital twins and validate the practical application.