Students develop a computer that can determine what an object is made of (and what it is)

Shawn Knight

Posts: 15,287   +192
Staff member

Students from the University of St. Andrews have developed a computer that can not only determine what an object is made of, it can also identify said object.

At its core is Project Soli, a tiny chip developed by Google’s Advanced Technologies and Projects (ATAP) group. The chip, which uses radar to detect subtle hand and finger movement, was designed as a unique way to interact with mobile devices.

While working with it, the students noticed that different materials produced unique signals. With a bit of machine learning, they were able to train the system to pull off some incredible tasks as seen in the clip above. They've dubbed their system RadarCat.

It’s not difficult to envision how this technology could one day be put to good use. For example, it could be used at a recycling facility to better help sort garbage or even help visually impaired people better navigate the environment around them.

Found is a TechSpot feature where we share clever, funny or otherwise interesting stuff from around the web.

Permalink to story.

 
The key elements would have to be (1) sensors and (2) the supporting data base(s) that have all the various compositions, chemical equations, etc depending upon just how sophisticated this thing wants to be. As stated, the students are "training" the system so it will be only as good as the inputted data, ie: GIGO!

Several years ago I read of a group of students that had built a mass spectrometer for around $400, which was astounding, especially after it was verified. Link that into this computer as well and you would have one very, very powerful device.
 
The only real use I can think of is forensics. But that requires detecting even tiny objects.

Not sure what else it could be good for.
 
I think that the only 'learning' that is happening is in the people that are developing it.
 
I don't see how testing the properties of things is any way a new idea.
This just looks like it is storing the data fed to it and recalling it based on the sensor input. What happens when it gets an object it has never been exposed to before, say a Bose Einstein condensate, or a glass eye.
 
This is brilliant! It's somebody seeing new uses for a pre-existing device, and the applications, with a little imagination, could be numerous. A robotic device with these little sensors in its hands could perform a lot of tasks, and would have, in some respects, a greater than human sensory ability. The database will continue to grow, and machine learning might even enable it to identify objects it hasn't encountered before, or at least give useful information about them.
 
Back