New research out of Germany holds promise for high-accuracy optical sorting of black plastics at reasonable costs.
Researchers at the Fraunhofer Society developed imaging and software technology capable of analyzing plastics of many different colors and determining the polymer.
“For the first time, we have developed an affordable sorting system that detects every plastic color, including black, both in real time and in large quantities,” Thomas Langle, department head at the Fraunhofer Institute of Optronics, System Technologies and Image Exploitation, stated in a press release.
The system, called BlackValue, is similar to existing optical sorters in that a conveyor belt carries and tosses plastic items off the end, using air jets to separate them. But instead of using a near-infrared imaging system, which has difficulty reading black plastics because they soak up much of the rays, the radar camera used by Fraunhofer researchers operates lower on the electromagnetic spectrum, closer to microwave and radio wave radiation.
The shredded plastic items come off the belt at a speed up to 10 feet per second. The camera emits a beam of radiation and, at the opposite end of the beam, the system analyzes how the individual plastic pieces have modified the spectra, allowing the computer to determine the plastic type.
A visible-light camera helps the computer understand the shape of the plastic piece so it can release a puff of air at the right time to send it to the right location.
Researchers also experimented with a camera operating in the terahertz frequency, a higher-energy frequency closer toward visible light. That higher-energy range provides for more precise measurements, but it’s much more expensive – possibly more than $1 million, according to the researchers.
So they looked for a cheaper combination that recycling centers could afford: a 90 gigahertz camera with less precise measuring capabilities. When paired with advanced software algorithms, however, the device is capable of detecting even minute differences in the spectra, according to the press release. The software is also self-learning, improving its accuracy over time.
The result is a 98 percent to 99 percent sorting accuracy and prices in line with hyperspectral cameras, according to the Fraunhofer Society.
The camera is expected to be on the market by the end of 2017. The technology was presented during the World Conference for Non-Destructive Testing in Munich, held June 13-17.
