The KIT spin-off Kinemic GmbH is developing a software for keying and device interaction based on steering gestures for industrial applications.
Dr Christoph Amma set off to become an entrepreneur with his doctoral thesis at the Chair of Cognitive Systems in Informatics at KIT: “While I was doing my doctorate, I explored options for recognising and interpreting human movements, especially gesture and handprint character recognition. Ultimately, this was to result in ‘airwriting’, a system enabling letters to be written and recognised in the air without a keying or input device. Texts are thus captured solely by movements of the hand.
Inventor Amma caused a sensation with this invention. His academic work earned him a number of distinctions, including the “Google Faculty Research Award” in 2013. Technical implementation was supported by Marcus Georgi, a fellow informatics scientist at KIT. The positive acclaim the two scientists have met with and the growing interest taken in airwriting have encouraged them to go on pursuing the development of the research results towards concrete applications. Thus the decision to go into business suggested itself. “Progressing from scientific proof to application at client level really is exciting,” Amma remarks. “You have to do a lot of developing to get your own vision put into practice.” Tomt Lenz, a fellow KIT graduate who joined the team, adds: “Thanks to my previous activities as a business consultant, I was able to provide valuable experience in company development.”
KIT’s spinoff emmtrix Technologies GmbH facilitates programming of multi-core processors and, thus, enhances performance of embedded computer systems.
What is already standard in desktop computers and laptops is increasingly applied in the area of embedded systems, such as telecommunication devices, automotive electronics or industrial control systems: Processors of two or more processor cores for higher speed and enhanced computing power. The higher performance of multi-core processors compared to single-core processors is only reached, if the tasks are distributed appropriately to several processor cores in an efficient and problem-free manner. So far, such a parallelization has been accomplished manually: This is very time-consuming, cost intensive and requires special programming knowledge.
A team of scientists of from the KIT Institute for Information Processing Technology (ITIV) headed by Professor Jürgen Becker started to look for solutions to simplify parallelization of multi-core processors already in 2011. Within the framework of the EU project “Algorithm Parallelization for Multi-core Architectures“ (ALMA), the scientists, in cooperation with industry partners, developed an innovative programming environment. The computer scientists and electrical engineers Dr.-Ing. Timo Stripf, Michael Rückauer and Oliver Oey were part of this research team. Based on the software tool developed in ALMA, they decided to start their own company. “About 40 person-years had been spent for the development. When the project was completed, we decided to start a company to make the technology available for the industry,” Timo Stripf explains.
Paralution Labs UG develops and sells its software PARALUTION, a software library for parallel calculation of sparsely populated linear equations. The software is hardware independent and thus gives inexperienced programmers the chance to use the newest research architectures. In addition to the development and sales of the software, the startup also offers services in the fields of “mathematics” and “parallel algorithms.” In an interview, we asked the team of Paralution Labs about the idea, the startup time, and the future prospects.