Multi-Sensor Recycling Sorter

Sorting Method

Belt speed 1.0×

Look-alike contamination rate 35%

Live Readout — active mode

Items sorted

Contamination events

Vision vs Multi-sensor — cumulative

Bin purity	👁 Vision	📡 NIR

👁 Contamination total

📡 Contamination total

Legend

PET bottle (clear, correct → Plastics) PVC look-alike (clear, correct → Reject) Aluminum can → Metal Cardboard box → Paper NIR scan beam reveals true material

How it works

PET and PVC bottles can be cut, molded and coloured to look identical on a belt — vision alone has no way to tell them apart, so a vision-only picker just guesses "clear bottle = PET" every time and routes every look-alike into the Plastics bin. When the item is actually PVC, that single wrong piece contaminates the whole plastics bale.

Near-infrared (NIR) spectroscopy shines light on the item and reads the spectrum reflected back — a fingerprint of its actual molecular bonds, not its shape or colour. A multi-sensor picker fires a quick NIR scan just before the pick, learns the item's true material regardless of what it looks like, and routes it to the correct stream.

Contamination rate — how often a clear bottle on the belt is secretly PVC instead of PET.
Bin purity — the share of each bin's contents that actually belongs there, tracked separately for each sorting method.
Cans and boxes look distinct enough that vision sorts them fine — the failure is specific to visually-identical, chemically-different pairs.

👁 VISION-ONLY

⚠ Contamination — wrong material in bin!

⚠ Couldn't load the 3D engineThree.js failed to load from the CDN. Check your connection and reload.