Sorting Technology Comparison | AISORT

Technology Capabilities at a Glance

Technology	Detects	Best Applications	Limitations
RGB Visible Camera	Color, brightness, shape	Plastic bottle color sorting, glass cullet, e-waste	Cannot distinguish same-color different-polymer types (e.g., clear PET vs clear PVC)
NIR Spectroscopy	Polymer type by molecular reflectance signature	PET/HDPE/PP/PVC separation; paper vs plastic identification	Dark or black materials absorb NIR signal; surface moisture causes spectral distortion
Hyperspectral / SWIR	Extended wavelength range for similar polymer discrimination	Dark plastic sorting, food-grade rPET purification, distinguishing HDPE from LDPE	Higher capital cost; slower scan rate than single-band NIR
X-Ray Transmission (XRT)	Atomic density differences between materials	Heavy metal recovery from shredder residue; mineral/ore sorting; removing aluminum from copper	Not suitable for light materials (plastics, paper); radiation safety compliance required
Eddy Current + Induction	Electrical conductivity of metals	Non-ferrous metal sorting (aluminum from copper); metal fragment detection in flake streams	Cannot identify polymer type, color, or non-metallic contaminants
AI / Deep Learning Vision	Visual patterns, brand-specific packaging, complex object geometry	Brand-level packaging identification; mixed-material component recognition; quality grading of sorted fractions	Requires representative training data; model retraining needed as packaging designs change

Matching Technology to Your Application

Rigid Plastic Containers (Bottles, Tubs, Trays)

Standard: RGB + NIR. RGB separates by color (clear vs. blue vs. green PET). NIR identifies polymer type (PET vs HDPE vs PP vs PVC). For food-grade output, add a second NIR pass plus metal detection to achieve <50 ppm contamination.

Flexible Packaging and Film

Standard: NIR + 3D laser. Film behaves differently on sorting chutes than rigid containers — it floats, folds, and overlaps. 3D laser triangulation helps distinguish film layers from rigid items; NIR identifies the polymer type of the film itself.

E-Waste and WEEE

Standard: XRT + RGB + induction + AI. The extreme density range in e-waste (from lightweight plastic casing to dense copper heat sinks and steel frames) requires density-based pre-separation (XRT), color-based sorting (RGB), and metal verification (induction). AI vision is increasingly essential for identifying specific component types like circuit boards, batteries, and connectors.

Construction and Demolition Waste

Standard: NIR + 3D + eddy current. C&D material is heavy, abrasive, and highly variable. Robust sensor housings with aggressive automatic cleaning systems are as critical as the sensor technology itself. Pre-screening to remove fines before optical sorting is essential.

Key Selection Criteria Beyond the Sensor

• Throughput vs. purity trade-off: For a given sorter width and sensor configuration, higher throughput reduces purity. Size your sorter for 70-80% of rated capacity at your design throughput to maintain adequate margin.
• Single-pass vs. multi-pass architecture: A single multi-sensor machine costs less but all sensors share the same material presentation. Separate machines in sequence cost more but each can be optimized independently — typically delivering 2-5 percentage points higher purity on challenging streams.
• Operating cost profile: Purchase price represents 40-50% of 5-year TCO. Sensor lamp replacement, valve rebuild kits, compressed air consumption, and calibration labor make up the balance. Eddy current and induction sensors have the lowest operating cost; XRT and hyperspectral have the highest.
• Service and support: A sorter is production-critical equipment — downtime costs can exceed the equipment cost within days. Verify local service presence, spare parts lead times, and remote diagnostic capability before selecting a manufacturer.