Funding provider: Funding from the Europäischen Fonds für Regionale Entwicklung under the Lower Saxony Multi-Fund Programme 2021-2027
Funding period: 1 May 2026 - 30 April 2028
Funding amount: €889,274.86
Project team: AI Methods in Process Engineering Working Group, Carsten Schilde
Summary of the overall project:
AI.Xplore is developing an AI infrastructure with real-world laboratories for robotics and drones to apply generative models in engineering research. At TU Braunschweig, AI.Xplore will create the infrastructure for a new, scalable AI innovation centre. The aim is to transfer modern generative AI technologies, especially Large Language Models (LLMs), multimodal models (LMMs), and Mixture-of-Experts (MoE) architectures, into engineering research, teaching, and industrial applications in a targeted way. The project centres on developing a GPU-based high-performance computing infrastructure for domain-specific AI modelling and establishing two application-oriented real-world laboratories with humanoid robots and autonomous drone systems. These laboratories provide a practice-oriented environment for testing intelligent, adaptive assistance systems in realistic engineering contexts, including process support, modelling and automation of technical workflows, materials handling, inspection, and digital quality assurance. In the long term, closed loops of experimentation, measurement, modelling, and hypothesis generation, also known as self-driving labs, will be established. By developing highly innovative AI-based demonstrators and real-world laboratories, the project lays the foundation for a new generation of interactive systems in engineering research.
Fig.: Advanced, hierarchical and modular Mixture of Experts Transformer (hMoET) network structure
Goals and tasks of iPAT:
Project partners: