Ryan J. Schermerhorn


Artificial intelligence (AI) is revolutionizing our world, changing how we live and work. The most pessimistic and fearful among us predict that AI will steal our jobs and, worse, take over the world. But at least in one area, AI is changing our world for the better. Indeed, AI is increasingly being used to help robots more efficiently and cost-effectively collect debris, separate recyclables from waste, and sort those recyclables by material (which in turn facilitates the recycling and reuse of the materials).

Since recycling became more common in the 1980s and 1990s, the burden of recycling items has fallen mainly on individuals who are asked to identify recyclable materials visually and then manually separate them from waste and into separate collection units. As a result, it is estimated that approximately 80% of items that could (and should) be recycled are not deposited into collection units for recyclables and are instead incinerated or directed to landfills. Additionally, items that cannot be recycled and should instead be incinerated or directed to landfills are frequently deposited into collection units for recyclables. Accordingly, when items that are collected for recycling are sent to a materials recovery facility (MRF), workers must manually remove these non-recyclable items. Workers must also manually sort the unique recyclable materials into different recycling streams (e.g., separate streams for plastic, paper, and metals), which is extremely time-consuming and labor-intensive. It’s estimated that the United States generates 300 million tons of MSW annually.

Enter companies like AMP and Glacier, which are developing innovative AI-powered technologies to address these problems. Other companies around the world are developing similar technologies, including, for example, Finland’s ZenRobotics, the UK-based Recycleye, and the French company Pellenc ST, but in this article, we will look at just two U.S.-based manufacturers.


In an effort to automate the recycling sorting process, AMP’s automated sorting solution is driven by at least four different technologies that appear to be covered by the manufacturer’s growing patent portfolio (currently at 23 families).

First, the sorting solution utilizes a modular vision system mountable to existing belts within an MRF. It features a patent-pending AI system trained to provide a real-time optical characterization of items as they are transported via the belts. AMP’s AI system can reportedly visually identify individual items by material type (including 50 different types of materials), form factor, and color at belt speeds up to 600 feet/minute and a belt coverage of approximately 80%.

Second, AMP’s sorting solution employs a patented belt-mounted vacuum system that includes an array of vacuum tubes mounted to the belts within the MRF that can reportedly collect up to 120 pieces of firm per minute. Once installed, AMP’s belt-mounted vacuum system is configured to remove thin films from the belts that the modular vision system identifies as having the potential to clog equipment and contaminate the recycling streams. AMP’s solution then utilizes one or more robotic systems and air conveyors to sort the remaining items into separate streams.

AMP’s robotic systems use several patented technologies to remove items, including suction grippers that can capture and hold items targeted by the modular vision system and a suction system that is pairable with the suction grippers (or other components) to facilitate removal. Meanwhile, AMP’s air conveyors feature several rows of air nozzles configured to direct air (and, in turn, items on the belts) into separate streams based on, for example, the weight or form factor of the items.


Glacier, a startup backed by Amazon’s climate tech venture fund, the Climate Pledge Fund, is piloting next-generation, AI-guided recycling robots to automate the recycling sorting process. As a first step, Glacier has developed a proprietary, custom-designed robot that is smaller (Glacier’s robots are human-sized) and reportedly cheaper than conventional recycling robots. Glacier has also developed a proprietary AI model that can apparently visually identify upwards of 30 different materials. When Glacier’s robots are equipped with Glacier’s AI model, the robots utilize the AI model to quickly and easily visually identify materials and automatically sort those different materials into separate streams, all while collecting real-time data about these different materials. Amazon estimates that “[o]ne Glacier robot can prevent over 10 million items per year from [incorrectly] ending up in landfills.

Patent Applications

Companies like these are also helping to increase the number of AI-related patent applications filed with the U.S. Patent and Trademark Office (USPTO). In 2020 alone, the USPTO published approximately 80,000 AI-related patent applications, significantly increasing over the approximately 10,000 AI-related patent applications published in 2000 (applications are published 18 months after their earliest filing date).

And while some of these AI-related patent applications may not be related to cleantech, we can safely assume that some are. Indeed, we know that in 2021, the last year for which comprehensive data is available, the USPTO received approximately 40,000 patent applications directed to cleantech technologies, which accounts for about 7% of total applications. We also know that AMP alone has approximately 100 AI-related cleantech patent applications filed with the USPTO, the European Patent Office, and the World Intellectual Property Organization (WIPO), which allows applicants to pursue international patent protection in approximately 150 countries. And because patent applications are not publicly available until 18 months after filing, it is quite likely that AMP has many more pending patent applications that have not yet published, e.g., those filed in 2023 and 2024 (we can also assume that Glacier has pending patent applications that have not yet published).

Climate Change Mitigation Pilot Program

To encourage this trend, in 2022, the USPTO introduced its Climate Change Mitigation Pilot Program, which allows patent applicants to expedite—at no cost—the examination of any U.S. patent application when one or more claims of those applications are directed to reducing greenhouse gas emissions. Due to the perceived success of the Program, the USPTO recently extended it for three more years (until 2027) and expanded the scope of the subject matter of eligible applications, notably to include claims directed to emission prevention and monitoring techniques.

Innovators are encouraged to leverage the USPTO’s Climate Change Mitigation Pilot Program to pursue patent protection for AI-powered technologies (and other technologies) that arguably address climate change in some capacity. This free program significantly expedites and lowers the cost of the patent application process.

We recently used the Climate Change Mitigation Pilot Program to expedite the examination of a client’s patent application directed to energy conversion. The patent application was filed in May 2023, and within just two months, the USPTO examined the application and issued a positive response, advancing all of the claims to allowance. The USPTO issued the corresponding patent in November 2023, just six months after filing (a process that traditionally takes two or three years).

Ryan J. Schermerhorn is a partner and registered patent attorney with the intellectual property law firm Marshall, Gerstein & Borun LLP. He may be reached at [email protected].
 DISCLAIMER: The information contained in this article is for informational purposes only and is not legal advice or a substitute for obtaining legal advice from an attorney. Views expressed are those of the author and are not to be attributed to Marshall, Gerstein & Borun LLP or any of its former, present, or future clients.
Photo by Igor Omilaev on Unsplash