The main aim of Google’s Project Soli is to use radar to add real-world gestures to smartphone and smartwatch navigation. Things like squeezing your phone to activate the camera or drumming fingers on the back to launch a given app are some examples of just how the technology could be used, but the University of St. Andrews in the UK came up with a different idea. Researchers at the University have put together a concept radar apparatus, called RadarCat, that is driven by machine learning and is able to recognize objects and materials.
RadarCat needs a host machine to run its software side on, while the hardware side relies on a specially made close-range radar built using technology from Google’s Project Soli alpha developer kit. The radar is incredibly precise, and with machine learning on board to help figure out what different material, shape, and mass compositions mean and how they go together, the technology is advanced enough to learn how to not only identify materials like metal and wood, or even to identify objects on a general basis, such as a phone or a laptop, it can be taught to identify specific objects by their proper name. In the YouTube video showing off the tech, for example, researchers place a Nexus 5 on the radar plate, and it is accurately identified. The much larger Nexus 10, which eclipses the plate entirely, gets the same treatment; this means that the waves are put out in such a way that they can wrap around and identify even larger objects. Fruits like oranges are also identified accurately.
One of the people involved in the project, University Chair of Human Computer Interaction Professor Aaron Quigley said that the university will be working toward not only implementing this new technology into wearables and other objects, but also toward testing the limits of the tech. The possible use cases and implications thereof are quite vast here; a future Soli smartphone could help a blind person to identify objects while shopping, the radar could be used for biometrics, or consumers could use the radar tech to check out a good they want to buy, such as a new laptop, to see if it’s the real deal or a back-alley knock off, among other uses.