Artificial intelligence: robots learn by imitation
Robots have long been indispensable in vehicle construction. They screw, paint, transport components and carry out many other activities. And as the automotive industry is now changing, these artificial helpers need to retrain. For instance, producing an electric car requires different actions than producing a conventional car does. , Head of the Institute of Intelligent Autonomous Systems, and his colleague Suman Pal want to make it easier for vehicle manufacturers to reorient themselves with artificial intelligence (AI). AI enables industrial robots to teach themselves new movement sequences – by mimicking a human model shown to robots in a video. “It saves a lot of time,” explains project coordinator Suman Pal, “because the time-consuming manual reprogramming is no longer necessary.” Computer science Professor Dr. Jan Peters
Professor Peters is one of the world's leading experts in the field of self-learning systems. He and his team have already taught robots how to play table tennis and how to juggle. Industrial robots in vehicle manufacturing have so far not been part of his group's research, but that is now going to change thanks to the support of the Pioneer Fund. With the funds now approved, the scientists want to demonstrate that their AI is suitable for the automotive industry. Ultimately, however, the concept is suitable for robots in a wide variety of industries, Pal points out, adding: “Our goal is to democratise robotics.” In the future, everyone should be able to reach a robot something – without any programming knowledge whatsoever.
Following the example of nature: precious metal-free catalysts for fuel cells
Although fuel cells are considered green technology, there is definitely room for improvement. To this day, they contain catalysts made of platinum – even though the precious metal is not only scarce and expensive, but often mined under questionable conditions. and her collaborator Markus Kübler from the Department of Chemistry deal with precious metal-free catalysts for fuel cells in their Pioneer Fund project. They are similar to our haemoglobin and, like the biomolecule, consist of carbon, nitrogen and iron. Other metals may also be involved in the nitrogen-carbon structures. TU Professor Dr. Ulrike Kramm
The TU researchers have already developed a manufacturing process for laboratory standard quantities of catalysts. Although extraction on a larger scale should not, in principle, be a problem, various reaction parameters would need to be adjusted for it. Kramm and Kübler wanted to address this as part of the Pioneer Fund project. Plus the focus is shifting increasingly to the use of the new catalysts. Among other things, long-term trials for stability are planned. They are to go well beyond previous experiments, which lasted only 24 hours or at the most a few days.
“With the support of the Pioneer Fund, we first want to look at the operating conditions that are used for the mobile power supply to smaller units,” says Professor Kramm. She cites fuel cell modules for energy supply when camping as an example, but points out that her research on electric mobility also benefits from this data. Kramm emphasises that she is always careful to “keep an eye on the big picture”. The Pioneer Fund is now helping her with this, because with the funds her team can deepen certain aspects and, at the same time, broaden the perspective.
ORGAN-iser: New system for 3D cell cultures
Pharmaceutical research is facing a fundamental problem: around 80 percent of the drug candidates that show the desired effect in preliminary experiments with cell cultures and animals go on to fail in human studies. This is where the Pioneer Fund project of of the Department of Biology comes in. Together with the groups of Professor Dr. Ulrike Nuber and Dr. Katrin Töpfer (Department of Mechanical Engineering) and Professor Dr. Steffen Hardt (Department of Electrical Engineering and Information Technology), they have developed a system called ORGAN-iser. Its purpose is the establishment of human 3D cell cultures that are similar to our organs and are to be used in drug testing. Professor Dr. Heinz Koeppl
A spatially fixed cell cluster is specifically supplied with nutrients and other additives from different directions in the ORGAN-iser. This is intended to improve the arrangement of the cells so they correspond to the structure of a real organ. “In conventional bioreactors, the cells do not develop as one would like. The spatial order is missing,” explains Dr. Töpfer, the project coordinator. The TU researchers work on 3D cultures from brain cells that they derive from human stem cells. The system is expected to be equally suitable for the development of other organ imitations.
Professor Nuber had a rough idea of ORGAN-iser in her mind before she joined TU Darmstadt five years ago. But it was only once she was here that she was able to realise the idea, she emphasises – and that was thanks to the cooperation of colleagues in engineering sciences, the expertise of the staff at the TU workshops and the support of the TU's Forum for Interdisciplinary Research (FiF). A prototype was produced in close cooperation. The TU researchers will now use the funds provided by the Pioneer Fund to validate their system. They also want to connect a number of ORGAN-isers in series in order to produce many organ imitations simultaneously.
The is an important contribution to the transfer of scientific knowledge from TU Darmstadt to business and society. Together with its partner Pioneer Fund, HIGHEST promotes innovations at a very early stage, increasing the innovative capacity of TU Darmstadt. Many of the supported projects will accompany HIGHEST at a later stage on the way to public funding such as EXIST or in spin-offs. ENTEGA NATURpur
HIGHEST manages the fund and coordinates all the activities throughout the application and funding phase. Funding decisions are taken by a commission made up of four representatives of Entega AG and TU Darmstadt.