Litcius/Paper detail

Deep Learning Enables Robust Drone-based UHF-RFID Localization in Warehouses

Chenglong Li, Emmeric Tanghe, Pieter Suanet, David Plets, Jeroen Hoebeke, Luc Martens, Eli De Poorter, Wout Joseph

20222022 3rd URSI Atlantic and Asia Pacific Radio Science Meeting (AT-AP-RASC)11 citationsDOIOpen Access PDF

Abstract

Radio frequency identification (RFID) localization technology has attracted great attention in stocktaking in ware- houses. In this paper, we investigate drone-based RFID localization for fast and accurate inventory management. Considering the drone trajectory errors, we propose a robust RFID lateral localization method based on the unwrapped phase, in which a temporal convolutional network (TCN) with non-causal convolutions is designed for the phase unwrapping. The tagged assets are localized via the nonlinear optimization upon the unwrapped phases. The experiment results in a logistic warehouse show that the proposed algorithm achieves RFID localization with 0.17-meter mean absolute errors and 0.4-meter 90-th percentile errors, respectively.

Topics & Concepts

DroneUltra high frequencyComputer scienceTrajectoryReal-time computingArtificial intelligenceRadio-frequency identificationPhase (matter)Computer visionTelecommunicationsComputer securityBiologyPhysicsGeneticsChemistryOrganic chemistryAstronomyRobotics and Sensor-Based LocalizationIndoor and Outdoor Localization TechnologiesAdvanced Manufacturing and Logistics Optimization