Skip navigation
Please use this identifier to cite or link to this item:
Title: Control and Processing of RFID Reader Array for Object Identification and Localization
Authors: Chazen, Zoe
Advisors: Verma, Naveen
Department: Electrical Engineering
Certificate Program: Robotics & Intelligent Systems Program
Class Year: 2019
Abstract: Recent developments in Artificial Intelligence (AI) have demonstrated the ability of machines to exceed humans in certain cognitive tasks [1]. Now, AI is extending beyond cyberspace to systems that are integrated into the physical world [2]. Examples include those focused on object identification and localization for perceiving human activity, a capability that would enable technology to better anticipate the needs of its users [3]. By preserving invariant structure in data, Physically-Integrated (PI) sensing, in which embedded signals are directly coupled with sensors, provides a solution to the challenges that arise when we seek to apply traditional ML models to dynamic and noisy real-world environments. Large-Area Electronics (LAE), which enables sensors to be distributed over large, flexible substrates, and Radio-Frequency Identification (RFID), which enables the identification and localization of tagged objects beyond line of sight, are two technologies that could enable a powerful PI system. This thesis designs, implements, and tests the CMOS-based control and processing portion of a hybrid LAE PI sensing system for object identification and localization based on an RFID reader array with RFID-tagged objects. By providing the missing link between LAE sensors and novel ML models designed to process their output, this thesis brings us one step closer to proving the viability of a PI sensing-enabled smart room for human activity detection. Once the RFID reader array and readout circuitry are completed and tested, this system will be integrated into a smart room equipped with other PI sensors, which include capacitive floor sensors and a wallpaper microphone array. The smart room will enable real-world testing of previously-simulated systems, whose ML algorithms infer human activity with greater accuracy and efficiency than traditional models.
Type of Material: Princeton University Senior Theses
Language: en
Appears in Collections:Electrical Engineering, 1932-2020

Files in This Item:
File Description SizeFormat 
CHAZEN-ZOE-THESIS.pdf8.74 MBAdobe PDF    Request a copy

Items in Dataspace are protected by copyright, with all rights reserved, unless otherwise indicated.