Fingerprinting Localization of RFID tags with Real-Time Performance-Assessment, using a Moving Robot

Abstract

This work is focused on unmanned inventorying and localization, by deploying an RFID-equipped autonomous robot. The robot is able to perform Simultaneous Localization and Mapping (SLAM), thanks to its optical sensors. As the robot moves inside the target area, it continuously interrogates all RFID tags within range. Passive RFID tags, placed at known locations, are used for the estimation of the locations of the target tags, by properly manipulating the measured backscattered power. The proposed method does not depend on the location of the reader, but only on the locations of the reference tags. Hence, positioning-errors related to SLAM are not accumulated. Mobility of the robot ensures rich collection of measurements. We propose a method for dynamic, real-time configuration of the parameters of the fingerprinting algorithm and real-time evaluation of the localization error of the unknown tags. This is achieved by treating the reference tags as target tags. Thanks to this property, we further exploit mobility of the robot, repeating inventorying and localization in areas, where poor performance is initially recorded. Measurements indicate a mean error of 18cm, with standard deviation of 11cm, deploying a single antenna.