Output details

Effective RFID based object tracking for manufacturing

This is the first research combining RFID with a probabilistic model-based architecture for manufacturing robot control. It was funded by the multi-million pound Auto-ID Centre (Mark Harrison, director, mgh12@cam.ac.uk), and partners including Procter and Gamble, Gillette, UCC and Cambridge-MIT Institute. It documents lessons from a five-year program to develop a full-scale RFID manufacturing demonstrator. Probabilistic modelling corrected 100% of the false positives (incorrect reads) and 95% of the false negatives (missed reads). Such correction is critical to the efficiency and correctness of automated RFID-based discrete manufacturing. The demonstration of approaches to resolve missed reads helped enable Wal-Mart to adopt RFID.