The Soli motion sensing system is a big selling point of the Pixel 4 series. In theory, it can quickly recognize user gestures and perform corresponding operations. However, the working principle behind this solution may not be as many people think, and Google explained it in a blog post recently.
First, Soli’s short-range radar is mainly used to do two things. One is to detect the presence of a person so that the facial recognition component is ready to scan the face. The other is to recognize the gaps such as sliding.
Empty gesture, This second item is the core of Soli’s functionality, and to make the relevant components small enough to be tucked into the phone, Google’s priority in radar recognition tendency is to put the action in front of the shape.
This approach also has the added benefit that Soli is relatively easy to violate user privacy. Because it has no way to record clear images. Google “does not have enough photos to recognize faces or bodies to use.”
And to capture the flashing motion of the user faster and more accurately in various usage scenarios. Soli uses a 3D view to observe the surrounding world.
It uses the vertical axis to represent the distance of the object, and the horizontal axis to represent the speed of the object approaching or leaving the phone.
The brightness of each pixel reflects the relative size of the detection object. The Soli sees when the user approaches the device, reaches for the device with his hand, and performs a sliding action on the device.
Google has built an AI model based on this and recruited a large number of volunteers to help Soli train gesture recognition with the TensorFlow machine learning framework.
The resulting results also need to be specifically optimized before they can be used directly on mobile phones by relying on Pixel 4’s custom low-power DSP.
Although in the current situation, Soli can only recognize some relatively simple actions, but in the eyes of Google, this technology has considerable potential to be tapped in the future.
Google believes that Soli is also expected to be integrated into smaller devices such as smartwatches in the future, and is, therefore, more suitable for devices operated by gestures. Also, it may also play more roles in security, entertainment, education, and other fields.