Researchers in South Korea and the US tested whether a laser can tell apart the four most common waste plastics (PP, PET, HDPE and LDPE) on real rubbish, not clean lab samples. After tuning how the data was processed, their model sorted every test piece correctly.
The technique, laser-induced breakdown spectroscopy (LIBS), fires a short laser pulse at a piece of plastic and reads the light it gives off. The team trained a simple machine-learning model on clean reference plastics, then tested it on separate pieces of real waste under deliberately imperfect conditions, closer to a working sorting line. Accuracy rose from 62% to 100% once they found the best of 23 ways to process the data and 7 ways to pick the most useful wavelengths.
What it means
Plastics that can’t be sorted cleanly are hard to recycle and often end up burned or landfilled, so faster, more reliable sorting matters. This is an early lab result: it covers four plastic types, the 100% figure comes from the best of many settings tried, and it has not yet been tested on a real sorting line. Worth watching, not a fix yet.
Nature (environmental sciences), 23 Sep 2026 · Peer-reviewed journal
Summary drafted from the paper’s published abstract and checked by an editor before publishing. Always read the original at the source link.