The display lights up when the hand approaches the watch, and a blue signal moves across the screen. Ivan Poupyrev rubs two fingers together. Song titles, news stories and the weather appear, and then disappear again. Mr. Poupyrev, a senior executive at Google, hasn’t even touched the device yet.
“We can revolutionize the interaction between man and machine with gesture recognition,” says the designer, speaking on a stage in Mountain View, California, where Google is headquartered. He calls it “a comfortable alternative to the current methods of control, through touch or voice recognition.”
At its internal I/O conference for developers, Google is unveiling a new language for the Internet of Things, one that could solve one of the industry’s central problems. Until now, the vision of an interconnected world, in which humans easily communicate with clocks, thermostats, loudspeakers or cars, has suffered because of usability issues. The Smartwatch, for example, a minicomputer that fits on your wrist, still looks like a miniaturized telephone with hardly any space on the screen for effective navigation.
But leading the way in tackling such issues is the Munich-based chipmaker Infineon, which provides the key hardware needed to make a user’s movements recognizable and transform them into signals for the watch’s operating system. DAX-listed Infineon has been working with Google as part of Project Soli since late 2015. The unusual partnership could redefine the way in which people interact with technology.
“For the first time in history, tools are oriented toward their users, instead of the other way around,” says Andreas Urschitz, who runs Infineon’s Power Management & Multimarket (PMM) division, which includes the mobile communications business. The innovative chip has great potential, says Mr. Urschitz, predicting that “it could turn into a market worth billions.”
Infineon expects to see a sharp rise in sales due to its business with smartphone makers.
In its partnership with Google, Infineon is undergoing a strategic turnaround. When Germany’s largest chipmaker sold its mobile phone division to Intel five years ago, it appeared to have turned its back on the mobile communications business for good. But instead, the company is back to supplying telephone manufacturers on a large scale. Its list of customers includes all major names in the industry, from Apple to Huawei to Samsung.
The semiconductor company also expects to see a sharp rise in sales due to its business with smartphone makers. “At this point, $2 worth of Infineon chips are in an average mobile phone. We assume that we can reach $3 in two years,” says Mr. Urschitz.
Infineon also expects to tap into the future market associated with the Internet of Things through its partnership with Google.
That’s because the chips with which Mr. Urschitz hopes to gain additional orders are completely new. One is a sensor that measures height down to the centimeter. This is unparalleled worldwide, Mr. Urschitz raved, explaining that it can be used for navigation in buildings, in addition to many popular fitness applications. Most current applications only recognize movement, but they do not detect whether something is moving up or down. “The chip reacts very sensitively to the change in air pressure,” Mr. Urschitz explained.
Much will depend on how well the chips actually work, though. In the case of Project Soli, the number of gestures and corresponding commands is limited to only about five. How smoothly users will interact with the technology also remains to be seen.
But even if the new developments are not as successful as planned, Mr. Urschitz is convinced that Infineon’s sales to mobile phone manufacturers will increase. “The new mobile communications standards require additional semiconductors in the devices,” he says. The fourth generation currently in use, better known as LTE, will be upgraded in the next few years. More and more semiconductors are needed, because all earlier standards also need to be met.
Still, Infineon is a long way from reviving its former, comprehensive mobile communications business. Until the beginning of the decade, the company was selling virtually all the internal components of a telephone. But the former subsidiary of electronics giant Siemens was finding it increasingly difficult to compete against significantly larger rivals like the U.S. company Qualcomm. This explains why then chief executive Peter Bauer was ecstatic when Intel paid more than €1 billion ($1.12 billion) for the division in early 2011.
Now Infineon is supplying components to mobile phone producers again, but only in selected areas where the company holds a leading position. This includes, in particular, chips for mobile phone chargers, as well as microphones and signal boosters.
PMM, Infineon’s second-largest division, last reported quarterly sales of €510 million ($572 million), or about a third of the company’s total sales. Automotive chips make up the largest share, with 42 percent of company sales. With last quarter’s operating margin of 15 percent, PMM was right on target in fulfilling current chief executive Reinhard Ploss’s yield objective.
It is important to Infineon that the mobile communications business does not depend solely on telephones, which is why the foray into so-called wearables, or wearable technology, makes sense. Sales figures are declining, now that almost everyone in industrialized countries already has a smartphone. According to experts with Juniper Research, only 320 million mobile phones were sold in the first quarter of 2016, a 6-percent decline over the same period last year.
Even Apple sold fewer iPhones at the beginning of the year then in the first quarter of 2015 – the first decline in the history of Apple smartphones. Although this put a strain on Infineon, other chipmakers were more seriously affected by the decline in mobile phone sales.
Infineon head Ploss remains optimistic and expects sales to be up by 12 percent in 2016.
The algorithms on data interpretation are from Google, while Infineon provides the sensors and the underlying chip technology.
The company is also engaged in developing next-generation technology at its Silicon Valley plant in the town of Milpitas as part of Project Soli.
Suresh Ram, who heads the section of the company that deals with radar and sensors, demonstrates how he is teaching computer chips to “see” and convert his movements into digital signals. “Look, if I move farther away now, the system can no longer detect me,” he says enthusiastically.
In his lab, Mr. Ram and his team are examining the interplay between software and hardware, as part of Project Soli. The company is betting on a new chip that will recognize gestures, thereby simplifying the control of mobile devices. Instead of the requiring the user to type onto the screen with his finger, Infineon’s technology converts hand and finger movements into commands.
“Wouldn’t it be great if the user could accept a call with a wave of the hand?” Mr. Ram asked rhetorically, waving his arm.
At a size of just 9 by 12.5 millimeters (0.35-0.49 inches), the chip, named BGT60TR24B, is especially well suited for use in mobile devices, where every millimeter and every gram counts. It is also intended to make manufacturers less dependent on the mobile phone, which is still used for most data operations today. “Companies can place the sensor where it actually belongs. Then the user won’t have to reach for the phone first, but will be able to interact directly with the device instead.”
To implement the project, Mr. Ram is in constant contact with joint venture partner Google. “We communicate as equals,” he says. “Everyone contributes in the area he knows best.” The algorithms on data interpretation are from Google, while Infineon provides the sensors and the underlying chip technology.
The most important principle here is that the technology conforms to the human, and not the other way around. “We try to understand which gestures users execute most frequently to express a specific command,” says the developer. For instance, rotating the wrist is universally understood as a signal to increase the volume, while moving the hand up and down indicates switching something on or off.
Infineon’s radar sensors convert each movement into digital signals and transmit them to the processor in the device at a rate of 4,000 images per second. In the processor, Google’s algorithms recognize the information within fragments of seconds, interpret it and send it to the operating system as a command for the watch.
Infineon chose radar technology for the data transmission sensors. “The process also works in the dark or through materials like clothing, it consumes less energy and yet it reacts with extreme sensitivity,” says engineer Ram.
The algorithm is designed to only interpret certain movements as commands, such as hand movements but not head shaking. The range can be adjusted to specific applications. The chip recognizes fine-motor gestures within a distance of up to 5 meters (16.4 feet), and up to 20 meters when it comes to rough gestures. The team was able to reduce energy use by a factor of 22. This is important to ensure that the technology doesn’t overload the batteries of mobile devices.
Britta Weddeling lives in Silicon Valley and also reports on the internet and technology industry. Joachim Hofer writes covers the high-tech and IT sectors from Handelsblatt’s Munich office. To ocnatct he authors: email@example.com and firstname.lastname@example.org