CoreLab - Projects

This AI-powered demonstrator is designed to enable research in Human-Robot Interaction (HRI). It provides a versatile platform for evaluating different instruction modalities like text, sketches or speech. The setup features a robotic manipulator and an object sorting station. The primary objective of this system is to enable complex sorting tasks through natural communication, completely eliminating the need for traditional manual programming. By utilizing AI-driven interpretation of user input, the platform allows even non-technical users to define workflows intuitively via a chosen natural instruction method.

Our air hockey system combines a physical table with an AI-controlled robot that has been trained using reinforcement learning. The training takes place initially in a realistic simulation before the strategies are applied to the real table.

A key area of research is the Sim2Real gap – the challenge of reliably transferring skills learnt in simulation to the real world.

This innovative screw sorter uses state-of-the-art AI technologies to precisely identify and sort screws using a camera. Through the use of unsupervised learning, the screws are automatically clustered based on their visual characteristics – without the need for any prior manual classification.

Research projects on developing an autonomous E-Scooter, as well as an accompanying simulation.

The SAFES project at Heilbronn University develops precise measurement methods and open-source energy models to capture the actual power consumption of AI hardware. By analyzing devices ranging from edge computing to GPU servers, the initiative aims to provide a reliable foundation for "Green AI" in industrial and automotive sectors. Supported by the Carl Zeiss Foundation, the project promotes sustainable technology development by making the environmental costs of AI transparent and manageable.

The autonomous maze uses AI to steer a ball through a maze by adjusting the tilt of the platform. It allows us to explore how reinforcement learning systems can learn through interaction within a simulated environment and how a trained policy can be transferred to a physical device.

A central area of research in this context is the Sim2Real Gap—the challenge of reliably transferring skills learned in simulation to real-world systems.

Our model vehicle is equipped with state-of-the-art sensor technology – including LiDAR, cameras and a wide range of other sensors. A high-performance GPU enables complex algorithms to be processed directly on board.

This system provides a safe and cost-effective platform for developing and testing autonomous driving technologies and driver assistance systems on a small scale – without any of the risks associated with real vehicles.