Sorting trash is a dirty, dull, and dangerous job. Recycling workers are more than twice as likely as other workers to be injured on the job, and stubbornly high fatality rates make refuse and recyclable material collection one of the nation's most dangerous occupations.
But with the rise of artificial intelligence, sophisticated trash-sorting robots are now turning up at recycling plants across the nation. Guided by cameras and computer systems trained to recognize specific objects, the robots' arms glide over moving conveyor belts until they reach their target. Suction cups or oversized tongs attached to the arms snag cans, glass, plastic containers, and other recyclable items out of the rubbish and flick them into nearby bins.
Glide, grab, flick: every second or so, on average, the arms identify a new target and pluck it out of the pile.
The robots — most of which have come online only within the past year — are just as accurate as human workers and up to twice as fast. With continued improvements in the bots' ability to spot and extract specific objects, they could become a formidable new force in the effort to divert tens of millions of tons of recyclable materials from landfills or incinerators every year.
"It's still very early days," said John Standish, technical director of the Association of Plastic Recyclers, a trade group in Washington, D.C. "But I'm very, very optimistic for the future."
Developed primarily by Denver-based AMP Robotics, Bulk Handling Systems of Eugene, Ore., and ZenRobotics of Helsinki, Finland, the robots are arriving at an opportune time. China, the world's leading importer of recyclable materials, sparked a global crisis last January when it stopped accepting some items due to environmental concerns and tightened its contamination standards for others.
Some recycling facilities in the U.S. closed, and mounds of waste piled up as exporters scrambled to find new buyers for bales of mixed paper, plastics, and other materials.
"There is always a demand for raw materials, but the recycled ones have to now be much cleaner and better sorted — closer to mined raw materials," ZenRobotics CEO Timo Taalas told NBC News MACH in an email. "Then the problem suddenly dissolves."
ZenRobotics has installed its AI recycling systems at two U.S. locations and in 10 other countries. Its largest robotic, called Heavy Picker, can lift 60-pound objects with its tong-tipped arm, making it especially useful for sorting construction debris.
On the front lines
Assembly lines at manufacturing plants have long featured robotic workers, of course. But until recently the "disassembly" of trash piles into their individual parts was a tough problem for AI systems to crack.
The problem is compounded when municipal recycling programs let residents toss all their garbage into one common bin — meaning that recycling facilities receive a jumble of dirty and unpredictable items that must be sorted onsite before they can be shipped out for reuse.
Several technology companies have devised trash-sorting solutions by pairing cameras and robotics with computer algorithms that use what's known as "deep learning" to improve garbage sorting. But "the leap was the computing speeds and artificial intelligence deep learning that allows the computer to actually look at a crumpled up bottle and recognize it as a Coca-Cola bottle," Standish said.
Matanya Horowitz, CEO of AMP Robotics, told MACH in an email that his company trains each Cortex robot by showing it thousands of examples of bottles, cans, packages, and other items. "It learns to identify all of these materials on its own," Horowitz said. "It learns to look for logos, shapes, and textures."
Horowitz said the Cortex system, now in use at three U.S. recycling plants, can pay for itself in three years or less — similar to the return on investment cited by ZenRobotics and Bulk Handling Systems.
A new awakening
"This has really been an awakening of the industry," said Steve Miller, CEO of Bulk Handling Systems, which makes a garbage-sorting robot called MAX-AI. The spidery-armed bot, which uses a suction cup as a grabber, has been deployed at three sites in the U.S. and three in Europe, with more of the bots on order.
Beyond taking over the worst jobs on the sorting lines, ZenRobotics' Taalas said, the increased efficiency may cut recycling costs and create more jobs at paper mills, plastic recyclers, and other firms that reuse raw materials.
And if the robots prove themselves, the environmental payoff could be even bigger: as more of our waste gets recycled and reused, less will end up in landfills.