Can AI Rescue Recycling?

Can AI Rescue Recycling?
Can AI Rescue Recycling?


Recyclers struggle with a shortage of workers and rising costs. Robots and optical sorters may be able to help make the business economically viable.

Recyclers across the U.S. are struggling, hurt by a shortage of workers and rising costs that too often make recycling uneconomic.

They are hoping artificial intelligence can help turn things around and boost recycling rates.

Recycling of municipal solid waste declined from nearly 35% in 2015 to around 32% in 2018, according to the latest figures from the Environmental Protection Agency.

How can AI help? By doing the sorting work in recycling facilities that a dwindling number of people want to do—and doing it better. AI-driven robots pick up recyclable trash at around 80 pieces a minute; people can sort around 50 to 80 pieces a minute. Optical sorters, a more established technology that’s growing more efficient thanks to improved AI, are much faster, sorting up to 1,000 pieces a minute.

Both options are being deployed by the country’s biggest recyclers, and many smaller ones, in a bid to increase the amount of material they can viably retrieve from the waste stream.

Filling the labor gap

It all starts with the labor issue. Sorting sites are often only 80% staffed and sometimes as little as 20%, says Cody Marshall, chief system optimization officer at the Recycling Partnership, a nonprofit supported by beverage, food and other companies that works with recyclers to improve their operations. Staffing shortages mean they can’t operate at full capacity. “AI can fill these gaps," he says.

It is helping to do that at the Boulder County Recycling Center, one of Colorado’s biggest recyclers. The facility brought robots into its sorting plant three years ago. It has two robots with arms and suction cups sorting plastic bottles, milk cartons and other recyclable trash on a conveyor belt. The robots are doing grueling jobs that few people want, says Suzanne Jones, executive director of Eco-Cycle, the nonprofit operator of the facility.

What’s more, “they don’t need breaks. They don’t go on vacation. They can work double shifts," Jones says.

And in the long run, sorting machines are cheaper than human labor, says J.D. Lindeberg, president of consulting firm Resource Recycling Systems. Recyclers typically recoup their investment in robotic systems in two years, he says. There are continuing costs for maintenance and upgrades, but robots still sort more for less.

Some recyclers are avoiding the upfront costs. Matanya Horowitz, chief executive of AMP Robotics, the biggest maker of recycling robots in the U.S., says many of the company’s robots are leased to customers, with fees that are 20% to 50% less than the hourly wages the recyclers are paying people.

Two forms of machines

As noted, AI-supported sorting machinery usually comes in two basic forms: robotic arms, whose popularity has surged in recent years, and optical sorting machines. Around 32% of the sorting centers in the U.S. are now using robotics, up from less than 5% in 2019, according to Resource Recycling Systems. Robots’ AI optical systems can scrutinize shapes, sizes and even brands through deep learning, a form of pattern recognition, to detect what is recyclable, such as plastic, paper, glass and metal.

Optical sorting machines are in most of the large facilities that together handle more than half of recyclables in the U.S., according to Resource Recycling Systems. They use sensors and lights to rapidly find what is recyclable on a conveyor belt of mixed materials. When recyclable materials are identified, the machines fire a burst of compressed air at them to sort them into bins.

Robots can be added to existing facilities, working on the same conveyor belts as humans, but optical sorters, which take in trash on their own conveyor belts, require additional space and typically are added in rebuilds or new construction of recycling centers, says Lindeberg of Resource Recycling Systems.

Big bets

WM, the biggest waste manager in the U.S., is betting on AI as part of its goal to boost its recovery of recyclable materials 60% by 2030. Last year, it started investing what will ultimately amount to more than $1 billion in recycling infrastructure including 40 recycling centers through 2026, with a big portion going to automation and AI.

An automated WM plant today might have four to six people sorting along with the machines, compared with up to 50 employees at a nonautomated facility, says Tara Hemmer, the company’s sustainability chief.

As one indicator of the effect of AI machines on WM’s operations, the company says that in recycling centers built around optical sorters, it has captured nearly 40% more polypropylene, commonly found in yogurt cups and butter containers.

WM’s automation drive is focused on optical sorters, but robotic arms are also making a difference in the sorting of polypropylene. The Recycling Partnership says it has seen a 259% increase in materials captured after robotics were installed at sites sorting polypropylene.

Republic Services, the U.S.’s second-biggest waste-and-recycling company, is investing in robots as part of its goal to recycle 40% more key materials by 2030, including cardboard, metal, paper and plastics.

The company plans to have robotics at around 20% of its 74 sorting centers by the end of next year, up from around 10% today, says Pete Keller, vice president of recycling and sustainability. Cutting labor costs can be the main reason to buy robots, depending on the location, but other factors such as how much material is recovered and its quality can play a bigger role, he says.

New challenges

Still, AI brings its own challenges. Robots require upfront spending and equipment that needs frequent maintenance and upgrades, Eco-Cycle’s Jones says. The cost for a single robot typically ranges from about $150,000 to $300,000, Lindeberg says. Building or upgrading a recycling center around optical sorters is even more expensive than robots and can require downtime, which some recyclers can’t afford. Optical sorting systems cost $1 million to $2 million each, Lindeberg says.

Minneapolis-based nonprofit Eureka Recycling opted to lease two robots in 2021 to sort bottles and cans instead of optical sorters because it couldn’t afford the downtime installing the optical sorters would require, says Kate Davenport, co-president of Eureka.

She says the robots are covering their costs with the value of materials they recover and have improved over time, but humans are still valuable in recycling facilities because there are still things they can do better. For one, a person can swiftly pull together a flood of plastic bottles coming down a conveyor belt, but robots have to use suction cups to individually pick up each bottle. “We see robots as one tool in the toolbox," she says.

There are also continuing costs for maintenance and repairs. But despite the fact that the recycler’s robots can break down for a total of up to two weeks a year, Davenport says, they are still paying off.

Horowitz, the AMP Robotics CEO, says upgrades and maintenance should grow less frequent in the coming years as the robots are fine-tuned. The learning curve is similar to that for robots in other industries, such as car manufacturing, he says.

Meanwhile, Lindeberg sees even broader potential benefits from AI-driven sorters, partly from the machines’ ability to record data on the material flowing through recycling centers. For instance, that data is increasingly helping some states learn about the brands in their waste streams, information they can use to create or expand so-called extended producer responsibility programs, which charge companies small fees on their products to fund recycling. Packaging designers can also use the data to help ensure that systems can recognize new materials entering the waste stream.

And AI can improve machines’ sorting capabilities over time through machine learning, he says. “AI technology’s impact on recycling economics has only just begun."

Dieter Holger is a writer in Santa Cruz, Calif. He can be reached at

Can AI Rescue Recycling?
View Full Image
Can AI Rescue Recycling?
Can AI Rescue Recycling?
View Full Image
Can AI Rescue Recycling?
Can AI Rescue Recycling?
View Full Image
Can AI Rescue Recycling?
Can AI Rescue Recycling?
View Full Image
Can AI Rescue Recycling?
Catch all the Business News, Market News, Breaking News Events and Latest News Updates on Live Mint. Download The Mint News App to get Daily Market Updates.


Switch to the Mint app for fast and personalized news - Get App