Using a Neural Network to Denoise a noisy RFID Signal in a Retail Environment
Ophelia Chan
A smarter, AI driven checkout method
In the pursuit of a smarter shopping centre experience for the forseeable future, one method that has been considered is the use of network connected RFID scanners integrated on the trolleys of a shopping centre to allow for an ‘on-the-go’ system where a customer can connect to their account, start shopping, and scan in items to their account to be paid for the moment they exit the bounds of the supermarket.
This project aims to create the basis of a neural network-powered AI decoding system for reading off chipless RFID tags, allowing for a cheaper and flexible alternative for smart shopping.
Page Views:
The overall purpose of this paper is to is to theorize the development of a low-cost version of this ‘just walk out’ shopping environment by utilizing RFID architecture in mobile units installed on the shopping trolley fleet of a retail outlet, instead of specialized display racks and surveillance, removing the need for purchasing, maintaining and storing expensive equipment, or having to retrofit an older retail outlet with new architecture to house and operate this new equipment. Reducing the setup to the trolleys and networking is likely to result in a drastic reduction to the overall costs of implementing a ‘smart retail’ system, as well as explore primitive examples of LSTM based neural network decoding of example chipless RFID signals from a range of provided noisy RFID signals.
The main benefits that are covered in this project involve the use of low-cost chipless RFID, combined with a neural network AI to create the foundations of an RFID based scanning system useable in a smart shopping experience
