Computers work at lightning speed and store more knowledge than any brain can. But when it comes to solving a task for which they are not specifically programmed, they soon reach their limits. Thus, even a beetle can find its bearings better in an unknown environment than a robot can.
Computers, however, are catching up. Since machines have been able to learn and thus have developed artificial intelligence, there have been systems that can reliably analyze images or speech and capture useful information for science and industry from large quantities of data.
And intelligent systems will be important in more and more areas of life: they could drive us as autonomous cars, help us out in the home on a daily basis; carry out emergency operations which are too dangerous for human, or diagnose and fight diseases in medicine as tiny robots. It's not surprising that both basic research and industry around the world are working to boost the intelligence of machines and that artificial intelligence
There are several locations and institutions in Germany – particularly n Baden-Württemberg – where the principles and applications of intelligent systems are already being researched on an international scale. Some of these activities are now being concentrated in Cyber Valley in the Stuttgart-Tübingen area. Funded by the state of Baden-Württemberg, this research network comprises the Tübingen and Stuttgart sites of the Max Planck Institute for Intelligent Systems, the University of Tübingen and the University of Stuttgart together with six industry partners: the BMW Group, Facebook, Daimler AG, Porsche AG, Robert Bosch GmbH and ZF Friedrichshafen AG. These core partners, especially the federal state of Baden-Württemberg, will be financing Cyber Valley in the next five years with a figure in the double digit million euros range. Moreover, NVIDIA, Trumpf and KIT are also interested in a partnership.
The aim of the cluster is to advance the research and development of intelligent systems, make this even more visible at an international level and in this way to attract the best minds in the area of artificial intelligence to Germany and the region. The research network also aims to support the training of junior scientists in this area.
In addition, the partners intend to use the new model of cooperation between science and industry to create a similarly stimulating environment for successful company startups in the area of artificial intelligence, emulating what Stanford University has done in Silicon Valley in the area of digital technology. After all, when it comes to the development of intelligent systems, the path from basic research to commercialisation is often very short. Startups that originate in the research environment are the engines of this development. The promotion of these startups calls for the close dovetailing of science and industry.
As core elements of Cyber Valley, initially five Max Planck Research Groups, funded by the key partners, will be established at the Max Planck Institute. Five further Groups will follow at a later point in time. Furthermore, in the first phase, four additional Groups, financed by endowment funds, will be set up at the universities. In addition, up to 10 new professorships, some of which will partially be financed by endowment funds, will be established at the universities. The new International Max Planck Research School (IMPRS) for Intelligent Systems will start in the summer of 2017. Over the coming six years, 100 doctoral students will be trained here. The newly created Cyber Valley groups will be accommodated at the two Max Planck Institute locations and at the universities before moving into their own new research premises on a permanent basis. In this way, a cluster with international visibility will develop in the area of intelligent systems.
The Endowed Chair at the University of Tübingen is funded by Robert Bosch GmbH. It will focus its research on machine learning methods, particularly the deep learning approach. With this method, software processes information – using a procedure inspired by the human brain – in hierarchically organized neural networks in which the level of abstraction increases from layer to layer. The crucial difference compared to other approaches in machine learning is that engineers do not specify how the level of abstraction increases from layer to layer. Instead, the hierarchical structure organizes itself based on the data using a universal learning method. In this way, programs learn in images using training examples to identify people or objects and to interpret entire visual scenes. Using such a deep learning algorithm, the “AlphaGo” software recently managed to even beat a top player in the board game Go, whose complexity computers had previously been unable to master.
Daimler AG will establish the Endowed Chair at the University of Stuttgart on the topic of entrepreneurship in an era of digital transformation.
The independent Research Groups that are being created in Cyber Valley will also be examining different ways of teaching machines to learn, particularly to make robots autonomous. As, in keeping with the principles of the Max Planck Society, the Leaders of these Groups are chosen for their scientific excellence and bring with them their own research topics, the specific research direction of the groups has not yet been established. However, Junior Research Group Leaders are not only sought in the area of machine learning and autonomous robotics but also in other fields of artificial intelligence. Another topic could be based on the way in which robots interact safely with people and still manage to perform a task reliably.
In addition, robotic perception is set to become a research topic in Cyber Valley: computer vision is the keyword here. The human sense of sight is still superior to computers, for example when it involves identifying objects in poor lighting conditions. This is not the only area in which computers will need to improve if they are to drive us autonomously some day, they will also need to learn to maintain a perspective in complex traffic situations. Such robots, which could one day help us as flexible household helps or even care assistants, must be able to analyze reliably the pixel data in their camera eyes.
In addition to mobility, household and care, medicine could also be an important area of application for robots. There will be demand in particular for microbots and nanorobots, which transport pharmaceutical substances specifically to the seat of disease in the human body, take samples there or even perform operations. However, research and development must make some progress before this can be achieved. Researchers in Cyber Valley will therefore also focus on robotics for medical applications.
Irrespective of the topics that the Research Groups and Endowed Chairs will focus on specifically, the declared goal of the cluster is to be able to quickly apply basic research findings. Cyber Valley will therefore drive technology transfer, for example by supporting researchers so that they can commercialise their findings in startups themselves. The Stuttgart-Tübingen region is destined in this way to become an important centre for artificial intelligence in science and industry.