Efficient and Accurate Motion Tracking and its Applications within the Internet of Things Design Space
Open Access
Author:
Gardner, Kevin
Area of Honors:
Computer Engineering
Degree:
Bachelor of Science
Document Type:
Thesis
Thesis Supervisors:
Mahanth Gowda, Thesis Supervisor John Morgan Sampson, Thesis Honors Advisor
Keywords:
ASL ASLLVD CNN HRI IMU LSTM Monocular RGB RGB-D RNN American Sign Language American Sign Language Large Video Dataset Convolutional Neural Network Human Robot Interaction Inertial Measurement Unit Long Short-Term Memory RNN Recurrent Neural Network
Abstract:
Finger motion tracking from unobtrusive inertial measurement unit (IMU) methods, such as smart rings and smart watches, is becoming increasingly useful as adoption of wearables become ubiquitous. Its applications range from better evaluations of a patient’s motor skills to real-time sign language translation. This paper explores two different methods for finger motion tracking—a hidden markov model (HMM) and a long short-term model (LSTM)—of the index finger metacarpophalangeal (MCP) and Interphalangeal Proximal (PIP) joints relative to the wrist using accelerometer data to predict relative position and discusses the tradeoffs of both methods. In order to overcome the challenge of limited IMU data, a synthetic IMU data preparation method was also explored.
