Date Thesis Awarded
5-2024
Access Type
Honors Thesis -- Access Restricted On-Campus Only
Degree Name
Bachelors of Science (BS)
Department
Computer Science
Advisor
Gang Zhou
Committee Members
Haipeng Chen
Ashley Gao
Huajie Shao
Abstract
Shoulder modeling is used to determine the position of somebody’s shoulder, but it is a complicated joint and difficult to measure, especially in an uncontrolled environment. Wearables on the subject’s shoulder could help monitor this without constant human surveillance, especially visibility is too limited for motion cameras to be effective. This is useful in a variety of situations; for example, when doctors use robots to provide necessities to a quarantined patient in order to avoid human contact and to monitor progress of shoulder injury recovery during physical therapy. Therefore, we created the ShoulDetector, made of a shirt equipped with conductive fabric sensors that changed their resistance in response to the shirt stretching due to shoulder movement. A detachable control patch has pin readings that measure the voltage, which is directly related to fabric sensor resistance. Using the motion detection software Kinovea as the ground truth, we use the voltage data to model the change in angle of the arm. Using data from 6 volunteers, we successfully used a polynomial regression and a Long-Short-Term-Memory (LSTM) model to detect the angles for the shoulder motions pitch, yaw, and roll. The results are accurate enough to determine the general position of someone’s arm.
Recommended Citation
Chen, Matthew; Koltermann, Kenneth; Zhou, Gang; and Shao, Huajie, "ShoulDetector: Shoulder Motion Detection with Conductive Fabric Sensors" (2024). Undergraduate Honors Theses. William & Mary. Paper 2087.
https://scholarworks.wm.edu/honorstheses/2087
