Recycling is broken. Can these robots help?

We're in the middle of a full-fledged international recycling crisis. The only hope may be to deal with the mess ourselves.

Robots step in to fix recycling mess We're in the middle of a full-fledged international recycling crisis. The only hope may be to deal with the mess ourselves.

Here's a fact that may strike you as bonkers: The complex where I live in a beach city adjacent to Los Angeles doesn't offer tenants a waste recycling option. For we steadfast few who insist on separating our plastics and papers, the only option is a trip to a local recycling center.

But I'm hardly alone. Turns out tens of millions of Americans don't have easy access to waste recycling programs. With recent news that China, which has been doing the lion's share of our recycling for decades, will no longer be accepting waste from the U.S., it's possible millions more will lose the service.

Worse still, recycling has never been all that efficient. "When the U.S. was sending much of its paper and plastic trash to China, for more than two decades, the bales were often so poorly sorted that they contained garbage," writes Adele Peters in a recent Fast Company piece. "The system never extracted the full value from those materials."

Robots, it turns out, could be a big part of the solution by upending the broken economics of recycling, which is why we started offshoring the job to China in the first place. Researchers at MIT/CSAIL, for example, have developed a robotic platform that can sort various items by touch. It uses sensors in its hands to determine the size and texture of waste materials like paper, plastic, and metal, and it sorts accordingly. 

Another company, ZenRobotics, combines data from multiple sensors along a waste stream to create an accurate real-time analysis. The robots use the sensor data to make autonomous decisions on how to grasp and sort objects. 

AMP Robotics is another example of a company combining robotics, machine vision, and AI to make recycling faster and cheaper, raising the possibility that we can onshore our waste disposal.

The system works much like a conventional automated pick-and-place system in a logistics warehouse. Waste items stream along a conveyor belt under the watchful eye of a camera. The system identifies items by type and a robotic picker with a vacuum end effector snatches items and flings them into appropriate bins.

But the easy explanation belies a huge challenge. Namely, garbage is gross. Things get crushed, making like items -- bottles, for example -- drastically different in size and shape. Food debris and liquids smudge or obscure labels and discolor packaging. Boxes are torn and food containers shredded.

The company solved the problem with a proprietary AI platform called AMP Neuron, which uses advanced computer vision and machine learning to train itself to identify items after processing millions of material images. The machine teaches itself to look for visual attributes like texture and size. Over time, and millions of cases of trial and error, it's become extremely accurate at identifying materials, committing identification errors in less than 2 percent of cases. That's more accurate than human counterparts, who work mind-numbingly long shifts bent over refuse. 

The machine vision approach also gives AMP the ability to capture data on the entire supply of refuse, information that can be fed back to recycling administers to determine how effective their programs are at collecting the right kinds of materials and sorting the right sorts of items.

Because the system is set up to learn from its environment, it can be applied to several kinds of recycling. For example, AMP is helping California-based electronics recycling company ERI. The robot in that case is performing 70 picks per minutes sorting through discarded disposed electronics. 

With increased efficiency and throughput, the economics of recycling begin to make sense. As cities around the country mull the future of their recycling programs, technology is offering a ray of hope.