E-Scrap Recycling: Efficient by Image Processing and LED Flashlight

Swedish Refind Technologies develops systems that automatically sort electronic waste and thus ensure greater sustainability. A sophisticated image processing system, which has access to millions of reference images, is responsible for the recognition of the respective objects. Thanks to digital image processing and LED strobe controllers, Refind was able to double the throughput of its machine.

The idea of the Swedish start-up team was to quantify the value of the waste before it is even recycled – perfect for recycling management with its slim profit margins. “But the actual implementation in the industry proved to be much more difficult. The batteries aren‘t lying on the table under laboratory conditions; rather, they come in highly diverse angles and conditions,“ says Farshid J. Harandi, head of mechatronics at Refind, explaining the challenges. A lot of research and development was necessary to complete the first model of the “Optical Battery Sorter“, which currently sorts eight batteries per second. The company from Göteborg, with six employees, is now generating around half a million euros in turnover annually, and it is the sole solution provider.

It is difficult to determine manually where the best returns can be achieved when recycling en masse. For example, certain models can be sold in emergent markets. “Here is where we come into play, because we can say exactly what is contained within a mobile phone and how much it may still be worth in another market. Then it can be sold at a completely different price than could be achieved by recycling,“ the head of mechatronics explains.


– Article continues after advertisement –

GR12015-14


The technology and algorithms behind the Refind systems are elaborate; the centerpiece is the image recognition and processing for controlling the machine in the sorting process. Batteries, for example, can come in many different forms: for torches, fire alarms or toys. Batteries can lie in the most diverse angles in the spectrum of 360 degrees; sometimes they are damaged or contaminated. No wonder that sorting is not a matter of child‘s play. “We have to be able to say what we are dealing with, with just this one glance. We have information about 2,500 types of battery and over two million images in the database,“ says Harandi, describing the process.

Light is especially decisive for automatically recognizing objects. “There was congestion with the old lighting control system, because when there were more batteries, there was less light exposure time. Therefore, we needed a stronger light,“ Farshid Harandi reports. The head of mechatronics browsed the market and, in doing so, came across the LED Strobe Controller IPSC2 from Smarek Vision, which, with its two channels, is the only device that operates with high-voltage pulses of up to 200 volts. “We can strobe the light with four times the intensity, using the Smarek strobe controller and, at 20 batteries per second, can now sort twice as many as before,“ says Harandi.

The machines sort half a million batteries per day, equating to a minimum of 600 kilograms per hour. The IPSC2 Controller actually even makes it possible to sort one metric ton per hour. However, this capacity is used to re-image problematic batteries. The clients are highly satisfied with this solution.

Source: Framos GmbH / Photo:  Framos GmbH

GR12015