“Trashbots” Boost Recycling at Pittsburgh Airport

There are more than eight million travelers passing through Pittsburgh International Airport each year, leaving behind some 2,000 tons of trash. Pittsburgh is a major American robotics hub, and now its airport has introduced an appropriately automated recycling solution: the “Trashbot.”

Trashbot uses cameras and sensors to scan discarded items. A machine learning algorithm based on Bayesian classifier sorts the trash for either landfill or recycle. If you dump an unfinished can of Coke, excess liquid will be drained off on a Teflon-coated sheet. The robot swallows garbage at the rate of around three seconds per item, with an 81% sorting accuracy rate, and continuously learns to recognize new types of trash. It is backed by a waste auditing dashboard based on Amazon Cloud Services, which presents data on types and quantities of refuse.

CleanRobotics CEO Charles Yhap and VP of Engineering Tanner Cook are social entrepreneurs. Cook regards trash as an important social issue, second only to energy, telling Synced that “due to contamination and confusion, only 20% of what goes into a recycling bin ends up being recycled.”

The team took two years to go from a wooden prototype at Pittsburgh’s AlphaLab Gear Startup Accelerator to product. Cook says the hardest part of building the startup was securing money to cover the initial R&D costs, not the 60-90 hours a week of work he put into the project.

What about cost? “Trashbot is comparable to what’s on the market, we’re looking at US$1,500 to 5,000 per can,” says Cook. For example, Denver international airport’s current trashcan cost is $3000, and Cook says his company could replace those for the same price. Clients also need to pay a small monthly subscription to use the data dashboard via Amazon Web Services.

CleanRobotics targets high-traffic locations such as airports, convention centers, malls, and schools. And although public facilities can be reluctant to adopt new technologies, Cook says many are actually reaching out to CleanRobotics.

2014 data shows a recycling rate of 35% in the United States, with numbers varying drastically across the country. California has the best recycling programs, and San Francisco tops the city list with an 80% success rate. Up north, Portland Oregon follows at 70%. The West Coast’s recycling records and proactive policies make the region an attractive market for AI-powered fast-track recycling solutions.

One of the biggest challenges facing urban recycling programs is the costs involved, particularly in sorting the trash to extract what’s worth recycling. A study on municipal recycling argues that recycling plastic and aluminum can actually be profitable for cities, while it is less cost-effective to recycle glass and paper.

Although it will be difficult to implement recycling procedures if they increase operating costs, many jurisdictions are offering tax credits to encourage recycling. Says Cook, “in Australia, it costs $350 per ton of landfill, versus getting a tax credit for recycling. When you have a facility that’s going through 10,000 tons of trash every year, it makes sense for them to try and divert as much of it into that high-quality recycling space as possible.” Trashbot is already onsite at a Sydney shopping mall.

The CleanRobotics team is continuing to use artificial intelligence and robotics to explore niche challenges and develop new product prototypes for waste management, and Cook believes the best is yet to come.