The machine, built by Claes Strannegard, an artificial intelligence researcher at the University of Gothenburg, in Sweden, has a camera and a computerized brain that runs on a neural network. That kind of system works more like a human brain in that it can learn to "see" patterns and respond to them. It's an ability that's important for sorting batteries, which come in a range of different sizes and shapes and contain materials, such as lead, cadmium and steel, that need to be distinguished from one another because they're valuable for resale.
At the recycling plant, batteries are fed to the machine on a conveyor belt. Its camera takes images of the batteries and its brain compares them to other batteries it has seen before. The machine may then send rechargeable "AA" batteries in one direction and single-use "AAA" batteries with steel casings in another direction.
The machine can recognize 2,000 different kinds of batteries and identify them in just milliseconds -- much faster than a human. And it can produce real-time information about how many batteries of a given type -- rechargeable or not, AAAs or Ds -- are being processed. This helps the recycling plant operator better manage the inventory that can eventually be resold.
The machine works differently from conventional mechanized sorters that scan for bar codes or color and are unable to discern a battery if it's dinged, dirty, dented or scuffed.
The battery-sorting machine was developed by Optisort, and so far, the company has delivered two machines -- one to Renova in Gothenburgand one to G & P Batteries in the U.K., which is sorting one-third of the country’s recycled batteries.
Maybe Skynet will be a sanitation worker rather than a general.
|