Key Takeaways
- RFID pre-read for three sorter types
- EPC encoding and routing decisions
- Exception handling and manual re-encoding
01RFID pre-read for three sorter types
The core of a sorter is "identify + divert". Traditional barcode scanning can't control parcel orientation and barcodes are easily damaged, with 85-90% read accuracy. RFID deploys a tunnel reader 2-3m upstream of the sorter, bulk-reading UHF tags on parcels (no line-of-sight, no orientation), binds the EPC to the destination and dispatches it to the sorter PLC, which diverts at the corresponding lane. A parcel hub lifted read accuracy from 88% to 99.98% and throughput by 35%.
Three main sorters fit different scenarios: cross-belt for small-medium parcels, high precision, high throughput (up to 36K/hr); popup roller for flat parcels, shoeboxes, envelopes; sliding shoe for large, heavy items (<30kg). RFID pre-read works for all three, completing identification+dispatch in 0.1s — far faster than the 0.5-1s mechanical divert, so it's never the bottleneck.
- Cross-belt: small-medium, up to 36K/hr
- Popup roller: flat, envelopes, shoeboxes
- Sliding shoe: large, heavy <30kg
- RFID identify+dispatch: 0.1s, no bottleneck
02EPC encoding and routing decisions
RFID sorting intelligence depends on EPC design. We recommend segmented encoding: first 24 bits = carrier+origin, next 24 = destination+delivery segment, last 48 = serial. After reading, the edge gateway parses only the first 48 bits to decide routing (which divert lane) — no central DB lookup, with latency stable under 50ms. One express carrier cut routing decision latency from 200ms (DB lookup) to 30ms (edge parse), lifting peak throughput 22%.
03Exception handling and manual re-encoding
RFID sorting isn't 100% lossless — 0.02-0.1% of parcels are unread due to tag damage, metal shielding or read-distance anomalies. These enter an "exception lane" where a clerk rescans or re-tags with a handheld, then returns them to the main line. One hub designed the exception lane as a "loop return" — exceptions auto-cycle back through the RFID tunnel for a second read; only those failing twice go to manual, cutting manual intervention another 70%. The centre's labour-to-machine ratio dropped from 1:8 to 1:25.
