From autonomous vehicles and aircraft through automated production to assistance robots and intelligent medical devices: intelligent systems for a sustainable society are the overarching vision of the University of Stuttgart. Autonomous systems are one of the research fi elds with enormous future potential.
The Cyber Valley innovation campus is one of the largest research collaborations in Europe in the fi eld of Artifi cial Intelligence (AI) and has become an international fl agship. The Max Planck Institute for Intelligent Systems, the universities of Stuttgart and Tübingen, the companies Amazon, BMW, Daimler, the automotive engineering service provider IAV, Porsche, Bosch and ZF Friedrichshafen are all participating in the initiative, which is sponsored by the state of Baden-Württemberg.
The research focuses on machine learning, robotics and computer vision. Ten new research groups and three university chairs have already been set up under the Cyber Valley program, and seven more have been put out to tender. The Cyber Valley is also home to the International Max Planck Research School for Intelligent Systems (IMPRS-IS), a postgraduate school for doctoral candidates. Last, but not least, the Cyber Valley provides an ideal environment for start-ups, breaking down the barriers between industrial research and basic research driven by pure curiosity. Researchers at the Cyber Valley are conducting cutting-edge research in their respective fi elds. They were recruited from all over the world in a targeted selection process to advance their research in the Stuttgart-Tübingen region. To provide some examples, we introduce a selection of them on the following pages.
Various research areas, supposedly as different as robotics, control engineering, computer vision, artificial intelligence and machine learning, all have one thing in common: they are all based on a cybernetic control loop. This is one of the research fields of Prof. Frank Allgöwer, head of the Institute for Systems Theory and Control Engineering at the University of Stuttgart and one of the co-founders of the Cyber Valley. the cybernetic cycle is based on the triad “perceive, learn, act”. If something goes wrong, instabilities can occur with sometimes dramatic consequences. Therefore, the ability to learn must be included in the algorithms.
With the help of the vacuum system visible in the background and a vapor deposition process developed by Prof. Peer Fischer's team at the University of Stuttgart’s Institute of Physical Chemistry (IPC) and at the Max Planck Institute for Intelligent Systems, it is possible to produce hundreds of billions of custom-made nanostructures ranging in size from 20 nm to 1 micrometer on a wafer in just a few hours. The computer-controlled growth process allows control over the three-dimensional shape and material composition. In this way, the scientists produce nano-propellers that they can move through tissue, among other things.
How can exciting scientific phenomena be viewed in virtual worlds and changed interactively and in real time? How can we humans influence the method of presentation or the visible data? This is what Michael Sedlmair, Junior Professor for Augmented and Virtual Reality at the University of Stuttgart’s Visualization Institute, wants to find out with his research. Together with students, he develops ideas and suitable algorithms that could enable such solutions and facilitate interactions between humans and data. One of the biggest challenges is to find out how complex data can be displayed in a real context and which new interaction possibilities are suitable for AR and VR technologies.
Prof. Marc Toussaint, head of the Machine Learning and Robotics group at the University of Stuttgart’s Institute of Parallel and Distributed Systems (IPVS) and a Fellow at the Max Planck Institute for Intelligent Systems, conducts research at the interfaces between artificial intelligence, robotics and machine learning. As part of the Cyber Valley project, he is working on how robots can manipulate and understand their physical environment to learn.
Robots often have to throw the towel in when sensitivity is required: they usually lack the sense of touch. Prof. Katherine J. Kuchenbecker wants to remedy this situation. The director of the “Haptic Intelligence” department at the Max Planck Institute for Intelligent Systems and her scientific colleagues are striving to equip robots with a sophisticated haptic perception. One project involves the “Intuitive da Vinci Si” surgical operating system. This robot-assisted surgical system already enables surgeons to perform a large number of minimally invasive procedures. In the future, the mechanical OR assistant should be able to provide haptic feedback.
One feature of AI-equipped systems is the ability to learn independently. The researchers working under the leadership of Prof. Ingo Steinwart, head of the University of Stuttgart’s Institute for Stochastics and Applications, rely on mathematical disciplines such as functional analysis, approximation and probability theory to gain an in-depth understanding of the methods and mechanisms of machine learning. This understanding is in turn the prerequisite for optimizing the existing learning algorithms of systems equipped with AI and being able to adapt them to new scenarios.
Industry 4.0 en miniature: the test and model plant for clocked production scenarios at the Institute for Control Engineering of Machine Tools and Manufacturing Units (ISW) is the model of a modern factory with digitally closed, flexible process chains. The plant, which was developed under the direction of Prof. Oliver Riedel, the head of the “Production Engineering Information Technologies” chair, does not only have actuators, sensors and control engineering at its disposal. An example: during operations, the workpiece carriers independently record measurement data, which are transmitted to a server with the aid of data analysis and AI preprocessing algorithms, where they are further evaluated. This allows production facilities to be monitored in real time, problems to be detected early on and workflows to be optimized - in the best case fully automatically.
Cyber-physical systems are becoming increasingly pervasive in automation technology. In this context, the Institute of Industrial Automation and Software Engineering (IAS) at the University of Stuttgart is working in collaboration with Siemens Corporate Technology in Munich, on digital twins as a virtual image of a physical system. The team, which is led by Prof. Michael Weyrich, concentrates on possible benefits of the use of digital twins in different application scenarios and created the demonstrator shown here for this purpose. Various aspects of the digital twin are made tangible using an illustrative example from the field of logistics. In the context of technology transfer, abstract concepts should thus find easier entry into industrial practice.
Perceiving, deciding and acting: For intelligent machines, this triad is the basis for being able to act independently in the physical world. The machines record the state of the environment via sensors, decide - based on this data - which action is to be carried out next and then convert it into an action. At the same time, they should learn from the data independently - for example, to become faster or better. Learning and decision-making abilities must be built into robots in the form of algorithms. Dr. Sebastian Trimpe, Cyber Valley research group leader at the Max Planck Institute for Intelligent Systems and guest scientist at the University of Stuttgart’s Institute for Systems Theory and Automatic Control (IST), and his team are conducting research in this field.