At this year's European Group for Intelligent Computing in Engineering (EG-ICE) conference in Glasgow and the European Conference on Computing in Construction (EC3) in Porto, five of our employees presented their current research projects. The event provided a unique platform for international encounters and knowledge exchange in construction informatics and engineering sciences. The projects presented dealt with highly relevant topics such as the generation of synthetic data for training AI models and concepts for data-driven building permits as well as data-driven construction logistics planning and control.

In ‘Leveraging Large Language Models for Synthetic Data Generation for AI Applications for Water Infrastructure Management,’ Jens Wala showed that OpenAI's large language models, especially the GPT-4 model family, can generate high-quality synthetic hydrological data. This data effectively addresses data scarcity in water management, improves AI forecasting accuracy, and aligns well with real-world patterns, providing a scalable and privacy-friendly solution.

Lars Wagenbach presented a concept for decentralised monitoring of rail traffic using extended reality (XR) with the aim of increasing the resilience of critical rail infrastructure. The current workflows of the control centres and their interaction options were transferred to an XR environment. The approach enables location-independent working and offers dispatchers an almost unlimited workspace.

Benedikt Kandler presented a framework for generating synthetic BCF data using LLM-driven multi-agent systems. The aim is to overcome data scarcity in the area of BIM issue tracking by simulating stakeholder communication.

Marcel Heiß presented a concept for the use of an extended form of the Information Delivery Specification (IDS) format in the context of BIM-supported, data-driven building permits. The geometric and semantic checks made possible by this allow for automated and transparent review of building applications using established standards, thereby increasing efficiency and traceability in the approval process.

Maximilian Gehring presented ‘ConLogAI – Concept for an AI-enabled platform for construction logistics scheduling,’ a concept for a BIM-based, AI-supported platform for construction logistics planning. The system uses BIM data as a basis, enriches it using AI-driven semantic methods, derives task dependencies and generates optimised execution plans using advanced scheduling algorithms. The goal is dynamic, resource-conscious scheduling that can adapt to disruptions in real time. Initial implementations demonstrate the feasibility of the approach and underscore its potential to significantly increase transparency, efficiency, and responsiveness in construction logistics management.

Attending the conferences not only promoted the exchange of new ideas and concepts, but also provided valuable impetus for future collaborations and projects. The participation of our researchers demonstrates the international relevance and innovative spirit of our institute, which contributes to shaping the future with a wide range of topics in construction informatics and engineering sciences.