Google Uses Millions Of Example Gestures To Enable Motion Sense On Pixel 4 Series

Google recently published an explanation post to shed a light on how the radar-powered Soli chip exactly works on the Pixel 4 series. In a word, this carries out the motion-sensing system on these latest Google’s smartphone models while enabling users to play/pause music with a swipe and so on. Truth be told, Google engineers have to work on a slew of deep learning algorithms to train Soli as an AI model.

Technically speaking, Soli’s short-range radar basically boasts the ability to recognize user presence and understand airborne tapping or swiping gestures. It’s safe to make small antennae for the smartphone, so the radar appears to detect motion rather than all the shapes.

According to Google, it also brings extra advantages in terms of privacy. Because Soli fails to form a qualified image, the manufacturer won’t get anything “distinguishable” from a user’s body or face.

As such, all Google wants to do is to interpret temporal motions as quickly and accurately as possible. For instance, the chip has to predict your movement while you’re walking at the same time the phone is playing music.

Under Soli’s eyes, the world is in the form of 3D graphs, with the vertical axis tied with distance and horizontal axis linking to the velocity of the device while the brightness of each pixel of the object results in its relative size.

Based on the method above, Google established an AI model and used millions of gestures to train the system. For now, Soli can only read simple gestures but Google reportedly aims to enhance the system in the near future. Aside from smartphones, the tech will soon dwell on other smaller gadgets like fitness trackers and smartwatches.

>>> How AI Powers A Lot Of Features On The Pixel 4