In Brazil, according to the World Bank’s What a Waste 2.0 report, the rate of collected plastic is 91% – not an ideal number, but worthy of praise. According to the International NPO WWF, the problem is that only 1.28% of the total is actually recycled and reinserted into the industry.

Bottlenecks between the collection of recyclable materials and actual recycling are many, but the solution forcefully touches one aspect: improvement of efficiency in the screening process of waste. The answer that comes first in the United States is investing in technology. After all, what’s more efficient than a smart high performance robot?

Speeding up recycling technology

In July 2017, China changed the rules of the international recycling game. In a meeting held by the World Trade Organization (WTO), in Geneva, the Chinese government announced that it would drastically reduce the import of solid waste, especially plastic and paper. Importing waste was a good business, but with environmental and social costs getting increasingly higher. 

Of all the world’s plastic waste, China received at least half. For paper waste worldwide, import went as up as 70%. United States, Germany, Japan and South Korea sent, each, to the Chinese territory between 250 million tons of waste and 1 billion tons in 2017 alone – in the same year, Brazil exported between 10 and 50 million tons.

“It was a huge shock for the global recycling industry,” said Arnaud Brunet of the Bureau of International Recycling, to German media company Deutsche Welle. After all, waste needs a destination. Southeastern Asian countries such as Malaysia, Thailand, Vietnam, Indonesia and India received part of the waste, but China’s message spread over the entire continent: no one else wanted to be the world’s landfill.

Humans out, machines in

The veto to waste export forced developed countries to create some strategy that solved, or at least minimized, the generation of large-scale waste. And urgently. The United States face the biggest challenge: they have a population of about 330 million people, with an average generation of nearly 1,200 kilos of waste a year per person.

The good news for Americans: of this immense amount of waste, 75% is recyclable. The recycling industry in the country is estimated in US$ 110 billion. That is, in addition to social and economic need, it is also an economic opportunity.

That is why companies and universities are in a fast race to develop robotics technology with cutting edge artificial intelligence to improve efficiency of recycling process. The goal is to replace human arms in collection and sorting of waste with tweezers and mechanized platforms, optical sensors and high processing computers.

Replacement of flesh and bone hands with machines is promising; in addition to increasing productivity, it lowers risks to the health of recycling workers. The work involving sorting and classification of waste usually takes places in dirty environments, often without proper protection. 

According to a report by the School of Public Health of the University of Illinois, these professionals are twice as much more susceptible to work accidents and injuries compared to the average of workers in the country – and they have higher fatality rates at work.

“They’re not going to replace people. We will absolutely keep quality control people in our facilities—but they will allow us to process more tons per hour and provide cleaner recyclable products,” said Steve Sargent, director of recycling for Rumpke Waste & Recycling in an interview with American publication Scientific American.

How does a recycling robot work?

An ordinary worker at a waste sorting conveyor can sort between 30 and 40 recyclable items a minute. Machines can multiply these numbers by hundreds.

Some robots are already operating in the United States and Canada. The first versions can screen about 70 to 80 items per minute with a precision rate higher than human’s. RoCycle, developed by the Massachusetts Institute of Technology (MIT), and Cortex, created by artificial intelligence company AMP Robotics, announced precision rates of about 90% at screening.

For RoCycle, MIT researchers created a simulation of human arms and hands equipped with a set of sensors. “Our robot’s sensorized skin provides haptic feedback that allows it to differentiate between a wide range of objects, from the rigid to the squishy,” explained Daniela Rus at the MIT. 

This sensibility is able to estimate item size and calibrate the force needed to hold it without causing damage to it and to the robot. Now, the challenge is to integrate the robot hands to its eyes, which means building a system that crosses haptic information with video data from the robot cameras.

AMP Robotics has taken this step. Cortex can sort 160 pieces per minute with two different operations. A surface system provides more rudimentary information on the object, but it is integrated to another system, a more sophisticated one, vision system, which promptly recognizes the pieces and activates the machine arms. “By using this technology you can increase the quality of the material and in some cases double or triple the resale value,” informed Mantaya Horowitz, CEO of AMP Robotics.

More impressive is the robot developed by Canada’s Machinex. The technology, which mixes an optical mixer and infrared light and a system of chains and magnets, sorts up to 3 thousand recyclable objects per minute – but fails when it comes to precision.

According to Scientific American, introducing artificial intelligence in the entire American recycling industry may increase gains in the sector by US$ 6.6 billion a year. 

This is more than enough to get optimistic: more technology and more money in favour of a more sustainable world.

Content published in September 2, 2019