Sometimes it takes an outsider to see a problem and realize a solution. While studying for her Master of Science in Technology Ventures at Carnegie Mellon University, in Pittsburgh, Pa., Sarika Bajaj says she was accidentally put on Intel’s experimental fashion team. There she learned that textile sorting and recycling were usually developed as segregated processes with companies specializing in one or the other. “That’s when I was introduced to this world of textiles and when I realized what could be done with the help of robotics.”
Her solution was to create Refiberd, a sorting and recycling system that uses artificial intelligence (AI) and robotics to turn post-consumer textile waste into thread.
Created with two cofounders—Tushita Gupta and Mingyue (Ida) Wang—Refiberd’s system “has the potential to eventually eliminate 93% of global textile waste,” the company claims.
Once the garments have been collected, they are subjected to the sorting process. First, they are sorted by color to retain as much dye as possible, and then they are shredded. This step eliminates the problem of garments comprising multiple layers, which cannot be easily sorted by sensors. The final phase is to sort the shredded material by fiber type, which is where AI gets to work.
“Here we use spectroscopy,” Bajaj says. “You shine a light on the material and that light reflects back, producing a graph of how it interacted with the material. Using spectroscopy alone doesn’t always correctly diagnose the material, but AI comes in handy because it can learn from the different graphs and approximate what the material is.”
Once the material has been sorted, it undergoes Refiberd’s patented chemical recycling procedure. So far, this is compatible with recycling polyester and cellulosics, and the company creates polyester and cellulose thread, which is compatible with existing manufacturing techniques to make new garments.