Whether it is in safety, data, reliability, consistency, handling of exceptionally dirty material streams, cost, or environmental impact, AI has helped bring a new level of performance into the realm of the possible.
By Dr. Matanya Horowitz and Carling Spelhaug
While the broader field of artificial intelligence (AI) has been around since the 1970s, the specific innovations we rely on in deep learning happened around 2013, driving major advancements in recycling, along with autonomous driving, facial recognition, and more, which began to become apparent around 2015. AI has proven adept at solving core challenges in the waste industry, and it guided our initial focus on deploying technology into recycling facilities with little change to their existing operations.
Since installing our first system in 2016, we have expanded our fleet to hundreds of automated sorting systems operating across three continents. We have fantastic partners, like Outagamie County in Wisconsin, who have leveraged our technology to address staffing shortages, recover recyclables more efficiently, and divert material from landfill. The facility has recycled enough aluminum, for example, to manufacture 16 million beer cans, according to Thomas Nelson, County Executive.
AI is one of, if not the most important tools making waves in the waste industry because of just how pervasive its effects are. Its significance can be seen in labor reduction, automation, the availability of data, recovery rates, purities, and more. Nearly every part of the supply chain is affected by it. While AI provides a notable efficiency boost when incorporated into existing facilities, it can be even more impactful when it is built in from the beginning at the facility level. It is this understanding that led us to introduce a facility design that is fully integrated with AI.
Retrofit Environments
When you build AI capabilities into a facility from the outset, you can optimize its design around the technology; a retrofit environment is limited by existing hardware, space constraints, and things like the speed and width of conveyor belts. This approach supports orders-of-magnitude changes in how AI-powered, next-generation facilities can perform. It is not just that you need fewer sorters, you can also operate without manual sorting altogether and add shifts more easily given the automation and reduction in variable costs. The deployment of vision systems throughout a facility provides the ability to monitor every line continuously. In retrofit environments, oftentimes, the most convenient places to install vision systems are residue lines or other areas at the front end of a system that tend to be high-burden locations, where material is piled on top of itself. When designing a facility with AI-enabled data collection in mind, you can make the lines wider and higher-speed—think more than 500 ft. per minute—than they otherwise would be to obtain the best spread and characterization data. AI can help replace entire classes of equipment that create maintenance and downtime events.
captures more material, and more value, from virtually any commodity or mix—all with minimal human
intervention.
Next Generation Facilities
While this all may sound like quite a departure from the status quo, recycling processing is already happening this way in a facility of ours outside of Cleveland. We are sorting mixed plastic and residue with minimal human intervention. The AI system that is operating the facility knows, in real time, the composition of what we are sorting, to get more value out of the waste stream. We can tag every bale we produce with a barcode that links to precise purity and composition information. Instead of a large team of sorters, we need only a small team focused on performance and maintenance.
This next-generation facility can take baled or loose material infeed; it is first run through a reducer to size it. Then, instead of humans, AI-powered equipment sorts the material. First, vacuum technology pulls off plastic film; then, the material runs through a series of high-volume air sortation jets that separate material by form factor, material type, color, polymer, and even brand—all using the same vision system as our robots do in retrofit environments. Double-cut sortation allows for the processing of two material streams simultaneously; each jet can be reconfigured in real time to sort for two different material streams. The system monitors the residue stream for heavies and purges when it reaches a certain level—all automatically. Fully autonomous quality control ensures quality without the need for human supervision. Having a vision system at every step of the process means it is possible to make intelligent decisions about the trade-offs between things like tons and purity to send as little value as possible to the compactor. These are only small examples of the capabilities of an AI-based system.
composition and purity of the materials that make it up.
Performance Benefits
Other performance benefits of an AI-controlled system include the ability to make macro-level decisions like how to recirculate material to ensure extremely low residue rates. In Cleveland, for example, our recirculation line gives us the option to have a second or third opportunity to recover more value from the residue stream. On a more tactical level, we can design jets that are self-cleaning and able to dislodge anything that might jam within them. With an optical sorter that jams less, it becomes realistic to think of design in a whole new way—a different framework for these facilities that means sorting dirtier material, handling more contaminants, and taking a different approach to the presort. To move the industry forward, we must design technology that is resilient to contamination and can more easily go after dirtier material streams, opening up pathways like multifamily, rural programs, and other domains where recycling penetration has struggled. It is the holistic benefits of an AI-controlled process from end to end that creates systemic improvements in the economics of a facility and makes it possible to leverage technology to bring new recycling programs online and drive regional improvements in recycling rates.
A New Level
The quickest path to wide scale change is aligning incentives with doing the right thing—in this case, that means sustainably managing our resources. If recycling is a better business for waste haulers, consumer packaged goods companies, and petrochemical companies, you get their alignment, and the industry can evolve quite quickly. AI and advanced technology can serve a range of stakeholders and make diversion and recovery as easy as possible. Whether it is in safety, data, reliability, consistency, handling of exceptionally dirty material streams, cost, or environmental impact, AI has helped bring a new level of performance into the realm of the possible. AI at facility scale will continue to transform waste and recycling and help to rapidly expand the scale and scope of the industry. | WA
Dr. Matanya Horowitz is the Founder and CEO of AMP. He developed and commercialized AMP’s breakthrough AI platform and robotics system, which automates the identification and sorting of recyclables from mixed material streams. He can be reached at [email protected].
Carling Spelhaug leads corporate communications for AMP. She has more than 15 years of public, private, and growth-stage company experience in public relations and communications. She can be reached at [email protected].