“How do you interact with an invisible computer?†If you hear such a problem in a conference room at a company like Google, readers will think that this will be a meeting to showcase future black technology, and there will be no physical display.
But in the Google conference room, a smart watch is placed on the table. When you hit a finger a few inches away from the smart watch, the dial will "react".
Ivan Poupyrev works at Google's Advanced Technology and Projects (ATAP) lab as the project leader for Project Soli. Project Soli aims to prove that electronic devices can be integrated with micro radar chips, and users can control electronic devices with gestures that are not very actionable. The radar is integrated into the smart watch to prove that the user can operate the invisible computer.
ATAP is an important part of Google. The last head of the company is Regina Dugan of the US Department of Defense's Advanced Planning and Research Bureau. The projects include modular mobile phones (Project Ara) and real-time 3D maps (Tango). ) and Spotlight Stories. Dukan moved to Facebook earlier this year, so whether or not the projects she left will continue is still an open question. Tango has "graduated" from ATAP and entered Google, and Ara seems to be in trouble.
But the Jacquard touch fiber project and Soli still remain in ATAP, and Soli has at least one new ambitious goal: to build a consumer electronics industry with radar technology and a design language. This is why the Poplev team not only conducts related tests, but also proves that radar can be integrated into smart watches.
Popplef said, "If a technology can be integrated into a smart watch, it can be integrated into any other product." ATAP redesigned the Soli chip, further reducing its size, reducing energy consumption, and more Optimize it a second time. According to Hakim Raja, Soli's chief and hardware product engineer, the team eventually achieved the minimization of the chip. It is very thin, and four antennas provide full-duplex communication and transmit and receive radar signals. The first-generation Soli chip in the development kit is consuming 1.2 watts, and the latest chip consumes only 0.054 watts, a 22-fold reduction.
There are drawbacks to making the chip so small. The radar is designed to detect metal objects flying beyond miles and beyond, rather than a few millimeters beyond the fingertips. Not long ago, people did not need to worry about this level of energy consumption, no one considered the impact of such a small radar chip on the signal.
Jaime Lien is a lead researcher at Soli, and her job is to optimize machine learning algorithms that are integrated into the chip. She first realized the meaning of translating the spatial signal provided by the radar into a time signal that can be processed on a computer. Nothing like the noise problem encountered by this small chip is more difficult. Her algorithm must find the required signal among many noises. It is impossible to bunch the signals, and each signal passing through the chip must be captured. In other words, this is a very complicated task.
Compared to electronic processes, the machine learning algorithm that determines how the electronic device reacts after the user gesture is relatively simple, but it is definitely not a pediatric. For devices with a touch screen, buttons and sliders are displayed on the screen; for devices with physical switches, the user can feel the scale when the switch is toggled. However, if there is nothing, how to guide the user to operate?
Jinhu Weibao Trading Co., Ltd , https://www.weibaoe-cigarette.com