3D gesture-recognition chip for smartwatches and HMDs

Researchers at the University of California at Berkeley and Davis are developing a chip that uses ultrasound waves to detect gestures. The chip is ideally suited to being used in wearable gadgets because of its convenience and low power consumption. Who wants to mess with a two inch smartwatch display anyhow.

Microsoft Kinect is an optical 3D processor for gesture recognition. It’s big and uses a lot of power. It can’t work in sunlight and is restricted in its general use. So, theoretically, these types of devices are not going to find widespread use in low power, small form factor devices including smartphones and tablets. The problem is almost insurmountable for smartwatches that have even a smaller footprint than phones, for example.

Richard Przybyla, a graduate student at UC Berkeley’s Berkeley Sensor & Actuator Center, the designer of the ultrasound gesture-recognition chip, called Chirp, has told MIT Technology Review that the small size of the touch screen on a wearable device limits what you can do. Przybyla’s Chirp team is building an ultrasound chip that doesn’t consume a lot of power and has a small enough footprint for smartwatches and head mounted displays (HMDs). 



Chirp uses sonar to detect hand gestures. Ultrasonic pulses from the chip echo off of objects. The chip can detect a range of gestures within a range of one meter.

According to the author of the article in MIT Technology Review:


Przybyla showed me a demo of Chirp at the lab he works out of at UC Berkeley, where the chips that comprise it were hooked up to a computer, allowing me to control a computer-animated plane’s flight path on a monitor by moving my hand in front of the display. Since the demo included a linear array of transducers, rather than a two-dimensional array, I was only able to check out Chirp in two dimensions (meaning I could control the plane’s side-to-side and forward-and-backward movements, but could not move it up and down). The group has built a chip with a two-dimensional array, but Przybyla says it’s still working to improve Chirp’s ability to track that up-and-down angle. It was noticeably easier to control on my first try than some other kinds of gesture-recognition technologies I’ve tried, and didn’t seem to require any calibration to sense most of my movements accurately.


Essentially, Chirp doesn’t need the same power consuming circuitry that a typical touch screen requires because it is sensing low-speed sound as opposed to light. The small screens on a smartwatch, for example, are annoying and don’t seem to advance user interaction in any way so, alternative methods have to be used to improved interaction on these devices.

Eventually, Chirp may be able to recognize more than hand gestures, providing feedback on whole body movement perhaps. The chips are about five millimeters across, but Przybyla thinks that they can go as small as one or two millimeters and still be effective for handling small hand gestures.