The economics of recycling are famously broken, but the math gets a lot simpler if the trash never gets mixed up in the first place. That is the quiet bet behind CleanRobotics, a Longmont, Colorado company that has spent the last nine years building a smarter trash can. Their flagship product, TrashBot, is a stainless-steel cabinet that uses cameras and a robotic arm to sort a tossed item into the correct internal bin before the human has even walked away [CleanRobotics]. It is a small, physical intervention aimed at a massive, systemic problem: contamination, which can render an entire load of recyclables worthless.
For a company founded in 2015, the pace has been deliberate. They have raised an estimated $10.8 million in total funding, including a $4.5 million Series A in 2022 and grants from entities like the U.S. Environmental Protection Agency [PitchBook]. The patience appears to be paying off in landing pads. TrashBot units are now installed in high-traffic, high-visibility facilities including Google campuses, Dallas Fort Worth International Airport, and Pittsburgh International Airport [SBIR, 2026]. The goal is not just to sort better, but to generate the clean data that makes waste management a predictable operational cost, rather than a guessing game.
A wedge of stainless steel and computer vision
CleanRobotics is not trying to build a mega-plant that sorts mountains of mixed waste. Instead, its wedge is the point of disposal itself. The TrashBot sits where people throw things away,airport concourses, office building lobbies, university cafeterias. Using computer vision and a simple robotic mechanism, it identifies an item as it falls and directs it into one of three or four internal streams: landfill, recycling, compost, or sometimes electronics [CleanRobotics].
The company claims the system can sort waste with up to 95% accuracy, a figure cited across multiple sources including the XPRIZE Foundation [XPRIZE Foundation]. For recyclables specifically, accuracy is reported at 90% [brio360.com, 2026]. In a world where single-stream recycling contamination rates can exceed 25%, that delta is the entire business case. The secondary product is the data dashboard, which tells facility managers exactly what is being thrown away and where, turning waste from a messy liability into a measurable, optimizable stream.
The team betting on first principles
The founding team brings a mix of hardware grit and academic rigor. CEO Charles Yhap is a serial social entrepreneur with two decades of experience focused on environmental and human rights issues [LinkedIn, 2026]. CTO and co-founder Tanner Cook handles the technical build. Perhaps the most distinctive credential comes from co-founder Koushil Sreenath, who was an assistant professor in mechanical engineering at Carnegie Mellon University, a background that suggests a first-principles approach to the robotics challenge [ri.cmu.edu].
This is not a team of waste industry veterans. Their bet seems to be that a fresh look at the fundamental physics and economics of sorting,applying modern sensors and software to a bin,can bypass decades of entrenched, inefficient practice. Their investor base reflects this hybrid of deep-tech and impact, including Monozukuri Ventures, Climate Capital, and the hardware-focused accelerator SOSV/HAX [Startup Intros].
Traction in regulated, brand-conscious facilities
CleanRobotics has found its early beachhead in a specific type of customer: large, regulated, and publicly visible institutions. Airports, universities, and corporate campuses face mounting regulatory pressure and brand incentives to hit zero-waste targets. They also have the capital budgets for CapEx-heavy solutions and the operational teams to care about the data.
The company now lists an impressive roster of flagship deployments, which serve as both revenue and reference sites:
| Customer | Facility Type |
|---|---|
| Corporate Campus | |
| Dallas Fort Worth International Airport | Airport |
| Pittsburgh International Airport | Airport |
| The Port Authority of NY & NJ | Transportation Hub |
| UNC Charlotte | University |
| AEG | Venue Operator |
These are accounts where a failure is highly visible, which puts a premium on reliability over pure sorting speed. CleanRobotics has also expanded its product line from the original TrashBot to include a slimmer, more economical model for offices and a larger "Zero" version capable of handling more waste streams [CleanRobotics].
Where the unit economics get real
The most credible challenge for CleanRobotics is not technological, but financial. The business model combines significant hardware costs with a recurring software fee for data analytics [GeekWire, 2018]. For a facility manager, the calculus is straightforward: does the avoided cost of contamination and the value of the waste stream data outweigh the lease or purchase price of the TrashBot, plus its maintenance?
- Capital intensity. Each TrashBot is a piece of engineered hardware with cameras, a robotic sorter, and software. This makes scaling manufacturing and maintaining field units a different game than pure software.
- Competitive landscape. They are not alone in applying AI to waste. Competitors like AMP Robotics and Glacier focus on the post-collection sorting at material recovery facilities (MRFs), a later stage in the chain with vastly higher throughput.
- The renewal motion. The long-term value is in the software and data layer. The risk is that the hardware becomes a low-margin vehicle for a SaaS product that is difficult to differentiate once the bin is placed.
The company's answer appears to be that point-of-disposal sorting is a defensible niche. By catching contamination before it pollutes an entire dumpster, they create immediate, measurable savings for their customers. The data then locks in the relationship by becoming essential for sustainability reporting and operational efficiency.
The next twelve months
With its Series A capital and a growing list of reference customers, CleanRobotics is positioned for a phase of scaled deployment. The next milestones to watch will be less about technical accuracy,95% is a hard ceiling to meaningfully improve,and more about commercial execution.
Key questions for the coming year include whether they can move beyond one-off flagship installations to multi-unit contracts with large facility management firms, and if they can drive down the unit cost of their hardware through design and manufacturing scale. Another signal will be the makeup of a potential Series B round; a lead investor from the waste management or industrial services sector would be a strong validation of their operational model.
On paper, the impact case is solid. If a TrashBot with 90% recycling accuracy replaces a conventional bin in a 500-person office, and each person generates 0.5 kg of recyclable waste per day, the bot could correctly divert over 45 tons of material per year that might otherwise be lost. That is the weight of about three city buses kept out of a landfill. The real test is whether facility managers will write checks for that outcome. To win, CleanRobotics must prove its bins are not just smarter, but more financially sensible, than the incumbent: the humble, dumb, and very cheap plastic bin.
Sources
- [CleanRobotics] TrashBot: The smart recycling bin that sorts at the point of disposal | https://cleanrobotics.com/trashbot/
- [PitchBook] CleanRobotics funding record | https://www.pitchbook.com/
- [SBIR, 2026] CleanRobotics project summary | https://www.sbir.gov/
- [XPRIZE Foundation] Reference to TrashBot accuracy | https://www.xprize.org/
- [brio360.com, 2026] Podcast mention of accuracy | https://brio360.com/podcast/
- [LinkedIn, 2026] Charles Yhap profile | https://www.linkedin.com/in/charles-yhap-83923a9/
- [ri.cmu.edu] Koushil Sreenath background | https://www.ri.cmu.edu/
- [Startup Intros] CleanRobotics funding and investors | https://startupintros.com/orgs/cleanrobotics
- [GeekWire, 2018] CleanRobotics' monthly software service | https://www.geekwire.com/
- [US EPA, 2026] Grant information | https://www.epa.gov/
- [AROptions, 2026] Customer deployment reference | https://www.aroptions.com/