Recycling is one of the easiest and most popular ways to pitch in and help protect the planet. However, figuring out what can and can’t be recycled is challenging at times.
For example, a greasy pizza box is cardboard. But, it’s not recyclable. Why? The grease actually taints the cardboard, thus making it unusable in the recycling process. Unfortunately, tossing one into the recycle bin will, in turn, contaminate even more items in the bin and possibly at a landfill. Therefore, making this kind of mistake kills the best eco-conscious intentions.
Thankfully, a group of MIT researchers at the Computer Science and Artificial Intelligence Lab developed a robot to tackle the pesky recycling contamination problem and the results might be surprising.
The Right Touch
Artificial intelligence (AI) is making many tasks easier in our lives. As such, Lillian Chin and her MIT research team aimed to leverage AI to find a solution to the global recycling contamination problem.
Her team developed RoCycle, a robot arm with grippers that can pick up items from a conveyor belt and identify what they are made of by touch.
The robot “uses capacitive sensors in its two pincers to sense the size and stiffness of the materials it handles.” By picking up items and “squeezing them,” RoCycle can tell the difference between paper, metal, and plastic objects. Once identified, the robot sorts the items accordingly.
The team created a mock recycling plant environment for the project. Objects passed RoCycle on a conveyor belt and it classified 27 objects correctly, with 85 percent accuracy. Thus, RoCycle had the right touch most of the time.
Benefits, Drawbacks, and Enhanced Visual Sorting
Overall, RoCycle’s success rate is impressive. Chin envisions robots like these could “carry out first-pass sorting” of recyclables at places like apartment complexes or college campuses. This would help to cut down on contaminated items coming into contact with other, “clean” recyclables. The result? Rather than millions of pounds of recyclables being reduced to waste because of contact contamination, whole batches of material could be saved and reused in future products.
When compared to robots that visually sort materials, the MIT team feels one benefit in using touch is that it’s more accurate.
“When you’re sorting through a huge stream of waste, there’s a lot of clutter and things get hidden from view,” said Chin. “You can’t really rely on image alone to tell you what’s going on.”
Chin has a point. However, one notable drawback is RoCycle can only pick up one item at a time. This makes the touch-based technology too slow for industrial recycling plants. These facilities have high operational costs and need to process waste fast to cover them.
In contrast, vision-based robots from ZenRobotics, a “leader in waste sorting robots,” sort 4,000 objects per hour.
Knowing this, the MIT team is aiming to combine the touch-based robot with a visual system to accelerate the process. In this scenario, a robot would scan all passing items but only pick up those it was unsure of.
Recycling Contamination is a Real Issue
According to the Environmental Protection Agency (EPA), people generated 262.4 million tons of municipal solid waste (MSW) in the U.S. in 2015. This translates to 4.48 pounds of MSW generated per person per day. Data also shows people recycled 67.8 million tons of MSW that same year.
The fact that much of this MSW might be contaminated is reason enough to applaud the work of MIT and the RoCycle. As the team continues to develop it, one can only hope it will become more efficient and effective in identifying contaminated recyclables, thus preventing masses of MSW from becoming contaminated by extension.