Subject Area
Computer Science
Abstract
In multi-user IoT environments collaborative quadruped robots, continuous user authentication is essential for safe operation. Current electronic leash systems for quadruped robots rely solely on possession of the remote control after initial setup, lacking multi-factor or continuous authentication. In this thesis by using robot observed human biometric gait pattern extracted by sensing user walking footstep that are identifiable from UWB-RF sensors signals used for follow companion feature, we developed a quaternion translated convolutional neural network (CNN) classification model that achieved 98\% accuracy in identifying authorized users based on user walking steps. Across multiple walking sessions, the binary classification model effectively performed user verification, validating the system’s capability for continuous user identification within 2-5 foot steps enabling collaborative robots to determine who they should follow and apply access policy. These mechanisms ensure accountability, prevent unauthorized access, and safeguard user privacy in collaborative autonomous robots, such as Unitree Go2 quadruped. A preliminary digital twin integration of Unitree go2 and the verification pipeline demonstrates the framework’s extensibility, enabling continuous authentication across both physical robot control and virtual monitoring scenarios.
Degree Date
Fall 12-20-2025
Document Type
Thesis
Degree Name
M.S.
Department
Computer Science
Advisor
Dr. Chen Wang
Acknowledgements
I wish to extend my deepest gratitude to my Advisor Dr. Chen Wang for his invaluable guidance, insightful feedback, and unwavering support throughout my research. I gratefully acknowledge the support of the Center for Digital and Human-Augmented Manufacturing (CDHAM) at the SMU Lyle School of Engineering for providing access to facilities, infrastructure, and collaborative research opportunities that greatly contributed to this thesis.
Number of Pages
85
Format
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License
Recommended Citation
Arumugam, Vinobalaji, "An unobtrusive RF Sensing Solution For Passive Human-Robot Interaction Verification" (2025). Computer Science and Engineering Theses and Dissertations. 52.
https://scholar.smu.edu/engineering_compsci_etds/52
