Many companies, especially in the healthcare sector, have so far refrained from using collaborative robots (cobots) because individual adjustments are costly and complex.
The KIROS project is developing an easily accessible training platform that allows cobots to be prepared for new tasks in a simulation environment without any programming knowledge. Users can interact with the platform via voice input and train their robots virtually. The skills learned are automatically transferred to real robots – without any manufacturer restrictions.
KIROS combines modern techniques such as reinforcement learning, domain randomisation and large language models (LLMs) to support small and medium-sized enterprises in autonomous automation. The project thus contributes to solving the shortage of skilled workers and to the widespread application of AI-supported systems.
Duration: 1 March 2025 – 29 February 2028
Projectpartners:
FUSE-AI GmbH
Helmut Schmidt University Hamburg, Chair of Electrical Machines and Drive Systems
Helmut Schmidt University Hamburg, Professorship of Computer Science in Mechanical Engineering
Letzte Änderung: 27. July 2025