After the sale ofBoston Dynamics to Softbank, it looked for a long time as if Alphabet's robotics division had been buried. But far from it, Google's tech forge X - Moonshot Company has been working on the current robotics technology for the manufacturing industry for five and a half years. The result: Intrinsic, a new, independent robotics company, has now been spun off from the creative tech pool. The new CEO is Wendy Tan-White, and the German robotics expert and former KIT professor Torsten Kröger has been hired as Intrinsic CTO; he has worked for Volkswagen, Kuka, Manz Automation, Redwood Robotics and Google, among others. According to Tan-White, the aim is to "unlock the creative and economic potential of industrial robotics for millions more companies, entrepreneurs and developers."

Robotics for electronics, automotive and healthcare

Tan-White bemoans "the manual and individual process that has been used to teach robots how to do things for decades, almost unchanged". She believes this severely restricts companies from exploiting the potential of robotics. "And many dexterous and delicate tasks, such as plugging in connectors or laying cables, are still not feasible for robots because they lack the sensors or software they need to understand their physical environment," continues Tan-White. Intrinsic wants to close this gap and make industrial robots usable for more companies, the manufacture of new products and innovative services.

The logo of the robotics spin-off.

© intrinsic

"We are leaving the rapid prototyping environment of the Moonshot Factory to focus on developing our product and validating our technology. We are currently looking for partners in the automotive, electronics and healthcare industries who are already using industrial robotics and want to learn together." Tan-White and her team write in great detail about the usual ambitious plans from Mountain View in several LinkedIn articles, statements and blog posts. Intrinsic is therefore focusing on the development of software tools that enable robots to be deployed faster and more cheaply and with less specialist knowledge. Tan-White compares the process with the lengthy and difficult construction of a website in the 1990s and the drag & drop construction kits of today.

German experts develop the technology

Robotics experts from Germany

In addition to CTO Torsten Kröger, German Engelberger Award winner Martin Hägele, robotics innovator Rainer Bischoff and "the luminary of reinforcement learning", Stefan Schaal, are already employed at Intrinsic. The young company is looking for robotics specialists through all channels, either for the Google headquarters in Mountain View, California or the development site in Munich. Jürgen Sturm is the new Robotics Lead in the Bavarian Intrinsic team. The tech expert writes in a LinkedIn article that he and his team are working on a solution that will make programming robots much easier. Together with the Alphabet subsidiary, he wants to develop "a simple and intuitive product that makes it easy for any user to perform tasks without limiting their potential to develop crazy ideas." Sturm is aggressively recruiting new employees with big American-style statements: "At Intrinsic, we have the exciting opportunity to bring groundbreaking technologies to real people and an exciting business potential in robotics."

A robot in the Intrinsic lab uses machine learning and force control to connect three different types of power plugs.

© intrinsic

Intrinsic wants to conquer industrial robotics with this technology

As CEO Wendy Tan-White writes in the presentation of the new robotics company, the team at X has spent the last few years researching how industrial robots first recognize their environment and continuously learn, optimize themselves accordingly and thus perform tasks even better. The aim is to make robotics "work in a wider range of environments and applications." In collaboration with several Alphabet teams and with partners in real production environments, the current Intrinsic team has tested software that uses techniques such as automated perception, deep learning, reinforcement learning, motion planning, simulation and force control.

ETH Zurich and Gramazio Kohler Research are partners in one of the first Intrinsic projects. Here, industrial robots are to assemble wooden components for an ecological construction project. In another test project, a robot was trained to correctly connect a USB cable within 120 minutes instead of hundreds of hours of programming. Simple tasks like these cannot yet be automated cost-effectively for companies. On this basis, Tan-White and her Intrinsic team want to revolutionize industrial robotics: "This shows the potential of the Intrinsic software to radically reduce the time, cost and complexity of using industrial robots - and thus help with a much higher number of problems in the long term and increase the variety of products that can be produced cost-effectively and sustainably."

The article first appeared on elektroniknet.de