AI / Autonomous tech / Environment / Robotics / Startups

Americans aren’t good at recycling. CleanRobotics wants to make it easier with AI

TrashBots automate waste sorting so you don't need to worry about what's recyclable or not. The startup was a top-10 finalists in the 2021 IBM Watson AI XPrize competition.

CleanRobotics' TrashBot. (Photo courtesy of CleanRobotics)

This editorial article is a part of Tech and the Environment Month of's editorial calendar.

A Pittsburgh startup is taking on the global recycling challenge with artificial intelligence.

CleanRobotics has a method patent for its core technology, which its leaders used to develop their TrashBot series — a collection of three smart trash cans that use artificial intelligence to sort waste for landfills, recycling or compost.

“If you’ve ever had the experience of trying to separate your waste — you know, landfill versus recycling — it’s just really confusing to this day, it’s still confusing to me,” CEO Charles Yhap told “And we got to thinking that we can build something that does this better and more accurately than people.”

Americans aren’t particularly good at recycling. According to the EPA, the national recycling rate in 2018 was only 32.1%. Pittsburgh’s numbers are worse: Despite a switch to single-stream recycling over a decade ago and education programs for residents, the recycling rate was last reported to be between 17% and 18% in 2018. Post-pandemic reports of those numbers will likely show even worse rates, as residential waste stream volumes have grown significantly throughout the past 18 months of stay-at-home orders and remote work.

But even those who want to be more intentional about recycling habits might have a hard time keeping up with varied rules and regulations across different cities, counties and states. What’s able to be recycled can change from office to restaurant to apartment complex to residential home, creating a challenge for even the best-intentioned recyclers.

If you've ever had the experience of trying to separate your waste — you know, landfill versus recycling — it's just really confusing. And we got to thinking that we can build something that does this better and more accurately than people.

“In many places, you can recycle a plastic bottle with a cap on it. And then in other places, that cap is not recyclable,” Yhap said. “So we’re able to calibrate artificial intelligence to recognizable features like that — little nuances — and make decisions based on on those features and those nuances and the local rules.”

Founded in 2015 by Yhap, VP of Engineering Tanner Cook and former CEO Vaish Krishnamurthy, the AlphaLab Gear startup accelerator alum was built in part by research led by additional cofounder and former CTO Koushil Sreenath, who worked as an assistant professor in mechanical engineering at Carnegie Mellon University at the time.

CleanRobotics’ three TrashBot models — the TrashBot, the TrashBot Slim and the TrashBot Zero — can each accommodate different volumes of waste and numbers of waste streams depending on what a school, office or other municipal and commercial customers might need. While the majority of people’s garbage is generated at home, Yhap and his team see the use of these automated technologies in nonresidential places as a way to focus recycling education efforts on residential settings, making the messaging around rules there clearer.

While the artificial intelligence and supporting architecture is complex, the process behind the TrashBots is simple. Passersby can get rid of waste by placing it into one of the bin’s openings, just like a normal trash can, but without having to read signs or figure out which opening is appropriate for their item. Instead, the receptacle takes care of that for them. Once the TrashBot receives a piece of garbage, it uses artificial intelligence to identify what the object is and whether or not it can be recycled, composted or sent to a landfill based on local regulations.

Charles Yhap. (Photo via LinkedIn)

The TrashBots recently helped propel CleanRobotics to being one of the top 10 finalists in the IBM Watson AI XPrize, for which fellow Pittsburgh startup Marinus Analytics won third place in June. In addition to bringing new press and credibility to the company, Yhap said that the distinction also helped CleanRobotics affirm the success of its technology: “As part of the competition, they were very rigorous in evaluating our technology, and making sure that it can do what we claimed it could do.”

Though CleanRobotics struggled like many other companies at the start of the pandemic to gauge future plans for the business, Yhap has witnessed a renewed interest in sustainable technology since businesses and offices started to reopen. (While CleanRobotics typically uses the East Liberty-based accelerator as its headquarters, the team has been working remotely since the start of the pandemic.) In addition to being more wary of the increasingly apparent downstream effects of climate change, he’s seen customers become more sensitive to costs associated with cleaning and waste removal in public spaces.

“Custodians are kind of the unsung heroes of the COVID-19 pandemic,” he said. “They have, depending on the place, four or five times the sanitation responsibilities compared to prior to COVID.” Those new hygiene protocols have companies looking for ways to save custodians’ time on the job by looking for ways to automate tasks, like waste management, that aren’t as closely related to public health and safety.

It makes the most sense to reduce waste everywhere possible. And to the extent that you can't, then at least it should be sorted as accurately as possible.

Beyond those cost optimizations, CleanRobotics has also benefited from more focus on waste reduction in general.

“Better diversion from landfill is kind of the core thing that gets people to the table with us,” Yhap said. “The beginning of the pandemic was kind of rough. But as things are opening up, we’re feeling really optimistic.”

CleanRobotics has plans to grow its team on both the technical and business side to support customer adoption and success as the company replaces old waste receptacles with its new models. Some new prototypes are also in the works, as are product pilots around the data collected by the TrashBots that could inform facilities on better waste management and supply purchasing policies in the future.

But despite having a business that relies on a steady waste stream for sorting, Yhap emphasized that the company’s mission centers on using their technology to create a more sustainable future.

“A lot of times people think that because there’s a commercial interest in what we’re building, that we want waste,” he said. “But you know, nothing could be further from the truth. It makes the most sense to reduce waste everywhere possible. And to the extent that you can’t, then at least it should be sorted as accurately as possible.”

Sophie Burkholder is a 2021-2022 corps member for Report for America, an initiative of The Groundtruth Project that pairs young journalists with local newsrooms. This position is supported by the Heinz Endowments.
Companies: AlphaLab Gear / CleanRobotics
Series: Tech and the Environment Month 2021

Knowledge is power!

Subscribe for free today and stay up to date with news and tips you need to grow your career and connect with our vibrant tech community.


What roles do gender and race play in the IT job market?

This Week in Jobs: Sketch out a new role with these 28 tech career opportunities

The importance of communication in MVP product design

'Talk to your customers all the time': Q&A with FoodSocial cofounder Bill Staley

Technically Media